Ghent Marathon 2026: follow the plan

The Ghent Marathon was my fourth marathon, but the first with a real dedicated preparation.

My first was in 2021 along the canal in Brussels. I had set out for a 30-kilometer long run and thought: I’m at 30 km, if I run 12 more I’ve reached my goal. I just did it like that, and finished in 3h45.

In 2025, I ran the Spa-Francorchamps marathon, but I had totally underestimated the elevation gain (damn you, Raidillon). Great experience, but I really blew up at the end and finished in 4h. A few months later, during my preparation for the Ultra Trail di Corsica, I had a 42-kilometer long run scheduled. I went back to the canal, but it was part of my trail prep so I didn’t go all out, I had targeted a pace of 5 minutes per kilometer and reached that goal.

In short, it was the first time I had the right conditions for a marathon. The first and third were solo along the canal, Spa was with elevation gain I hadn’t anticipated. The marathon is nothing like a 20 km or a 10 km, it’s truly a distance of its own. I was eager to see what I could do with a real preparation.

The preparation

This year, I started coaching with the Running Academy. Before that I was using Runna, but it’s an app a bit like Campus Coach, not specific enough. On top of that, since I also have big trail running goals, it didn’t cover my needs. I love running and I wanted to improve my training, and I was recommended Matthis from the Running Academy. We started working together in 2025 and he’s the one who built my preparation.

On average I was running five or six times per week, with the biggest weeks at 100 km. The specific preparation really started in January, after the Corrida Brussels (a 10 km race) in December. From there, we switched to a three-month marathon block.

As you can see on the chart, I kept doing CrossFit throughout the preparation. Honestly, CrossFit probably helps very little for marathon performance, and I think if I had skipped those sessions I would have run a better time. But I do it because I enjoy it and because I want to progress in that discipline too.

Shoe-wise, I wore my Adidas Boston 13 on all tempo runs and long runs during the last month of preparation, roughly 100-110 km. I had experimented with the lacing to avoid pain, and it was good to have tested them before race day. The important thing is to have them well adapted to your feet to avoid friction.

The weight dilemma

I’m 1m88 and 91 kg, not exactly your typical marathon runner build. During the prep I wanted to improve and I thought: I’m going to try to lose weight. There’s this rule that says one kilo less equals 5 seconds faster per kilometer.

I lost about 5 kilos, but at that point I was really weak at CrossFit, I couldn’t do the exercises anymore, I didn’t feel well. On top of that, flu-like symptoms hit. At first I stressed because I didn’t know I was sick. I was super tired, sometimes instead of getting up to go run I would just sleep. I thought: I can’t keep up anymore, it’s too much, I’m overtrained, I’ve asked too much of my body.

When I learned I had this minor illness, it actually relieved me. My routine was there, it wasn’t overtraining. Since I was weakened, I decided to stop trying to lose weight. Rather keep my natural weight, keep training to gain muscle and lose fat, but without cutting calories. Keeping the energy was the priority.

The diet itself wasn’t that hard mentally, I wasn’t particularly hungry. However, even though I was lighter, more dynamic, more agile on the run, there was a negative effect: with less energy, I felt less performant.

The goal

At the Corrida Brussels in December, I was targeting 40 minutes for the 10 km. I went out a bit too fast on the first kilometer so I didn’t hold, but I was close. From there, I could estimate my MAS at 16.7 km/h, which pointed to a potential marathon pace of 4’30.

In training, Matthis saw that 4’30 was a bit too fast. MAS gives you a range, but by looking at heart rate and training sessions, we refined it. We settled on 4’40 as the target pace.

Ghent, Sunday March 29th

I arrived in the morning, the race was great. The weather was beautiful but cool, around 5-6 degrees. I went to a thrift store to buy a pullover to wear before the start and abandon at the starting line.

I hesitated quite a bit about the outfit. I always run in a singlet, but with my build I tend to heat up very fast. I considered wearing arm sleeves, but I figured I’d end up taking them off. I went without and it turned out fine.

Kilometers 1-20: easy

The course is superb. The first 20 km, honestly, it’s easy. I was very careful to follow my 4’40 pace, because I knew that if I went a bit faster I could cross into threshold territory, and that’s not ideal. I really didn’t want to get carried away: at the 10 km race, I had gone a few seconds too fast on the first kilometer and it had put me in trouble.

Something that really helped was a trick I saw on Instagram: adding the average pace per kilometer on my watch, in addition to the instantaneous pace. The instantaneous pace jumps around because of GPS bugs. The average pace is much simpler to regulate with. If I’m at 4’35 average with 200 meters left, I know I can ease off a bit. It allowed me to be a metronome.

I also drafted behind certain runners who were holding a steady 4’40. You had to be careful because sometimes they’d speed up. When I saw on my watch that the guy was going 4’30, 4’25, I’d let him go. But there was one I followed for 8-10 kilometers, it was perfect: I tucked right behind him and it shielded me from the wind.

The Ghent course is really beautiful. You pass through small areas where there are lots of people cheering, sometimes in narrow passages with people right next to you, it’s motivating. And there are sections on bike paths along the canals, long straight lines stretching for kilometers. Beautiful in nature, but you really feel the wind. Drafting helped a lot on those exposed sections.

Kilometers 20-30: it burns

Between the 20th and the 30th kilometer, the muscles started to burn. Glutes, hamstrings, and toward the end the shoulders too. I was trying to relax a bit.

The shoulders surprised me because I’d never experienced that before. I think I wasn’t relaxed enough in my upper body. It’s a bit like the feeling in Hyrox when you do the farmer carry, holding weights at the end of your arms. I thought: « what’s going on, I’m not carrying 20 kilos in each hand, why is this happening? »

Kilometers 30-42: the paradox

The last 10 kilometers, there’s a paradoxical effect. On one hand it’s clearly the hardest part of the marathon. But on the other, when you start seeing the number 3 on your kilometer counter, you know there are only 10 left. There’s a real boost of motivation.

Nutrition-wise, I had planned 45 grams of carbohydrates every 7 kilometers, roughly every half hour, plus a gel right before the start. My very last gel, arguably the most important one, was at the 35th kilometer. I took it from my belt and it slipped out of my hands. I wasn’t quite lucid anymore I think. I thought « oh no », but I kept going, I didn’t stop. I ran the last 7 kilometers without any fuel.

It was a mistake because at the aid station there were candy, gels, fruit compotes. But I was too focused, I hadn’t noticed, I didn’t want to lose time. That’s part of the race day surprises. On top of that my gels didn’t contain electrolytes, and I got a small cramp. Something I’ll pay attention to next time.

The lack of carbs at the end definitely played a role. The last 2-3 kilometers I slowed down, ran them at 4’50, 4’55, the very last one at 5 minutes. But since I had banked a little time earlier, it balanced out and I maintained an average of 4’40 on the dot. It’s pretty funny to land exactly on the target number.

Around kilometer 37, I was really hot so I decided to take off my singlet. It’s something I always keep as a boost: the wind evaporates the sweat and gives me better thermoregulation. Obviously the last kilometers of a marathon are the perfect moment for that. And since my build stands out a bit from the typical marathon runner, quite a few people in the crowd started cheering me on with things like « let’s go beast! ». That was pretty funny.

In the final kilometers, I also followed another runner who started talking to me in Flemish. I didn’t understand a word and told him « English or French? ». We ended up chatting a bit during the race: we agreed to hold the pace until kilometer 37 and then try to pick it up. He was targeting 3h15.

The finish: the indoor stadium

At the finish we were together. The Ghent Marathon ends inside an indoor stadium: you enter, run a lap on a 400-meter track with the crowd in the stands. It’s absolutely magnificent, you’re really fired up. I remember I started semi-sprinting, passed a guy in the final meters. The atmosphere was truly electric, exhilarating. My companion from the last kilometers finished in 3h15min30s, he was happy.

My friend Martin was there too, we had dinner together the night before. He lives in Ghent and was running the half marathon starting at 1:30pm. I arrived around 12:20, so we had time to debrief before his start.

3h17’31"

Honestly, I went in not really knowing what to expect. I followed the plan, but with the minor illness and a rough patch at the end of the prep, I wasn’t sure how it would go. I told myself: just go and do the work. And it worked out.

The prep was done. The long runs were in the legs, the 20 km at marathon pace in the middle of a 90+ kilometer week were done. I went in without overthinking it, and race day went well. A good preparation, I think that changes everything. The marathon doesn’t forgive, and when the conditions aren’t right, it usually doesn’t pay off.

I absorbed the training load well. When you have a bad marathon it passes extremely slowly, and this one went by fast, even though the last kilometers were super long. I was well within my paces, I hadn’t gone out too fast. Physically and mentally, you recover quickly when it goes well.

I can see there are still areas for improvement, and that’s motivating. I’m starting to know myself well as a runner, and it’s important to appreciate these milestones.

I still love it just as much, that’s for sure. If I had to give one piece of advice to someone preparing for their marathon: enjoy what you’re doing. That’s really the point, have fun with it.

Thanks to Matthis for the coaching and the preparation that made all of this possible.

What’s next

The next big goal is the 70 km by UTMB in Romania, end of July. Until then there will be a few smaller races in B-race mode: a Hyrox and a CrossFit competition, which will be my first.

I also have a road race date at the end of May, the 20 km de Bruxelles. It’s the most popular race in the city, almost 30,000 participants. The idea is to fit it into the trail preparation while keeping one speed session per week. It’s important not to lose what’s been built, but the priority will shift to trail and elevation work.

Ghent was a great marathon, not too much elevation. I like running close to home, I live in Brussels, and it’s nice to take the train rather than a plane. But I like variety: there are other marathons not far away that I haven’t tried yet, Namur, Paris maybe. I’d happily do Ghent again, but I’ll probably go discover another one.

I’m thinking of doing the same kind of program next year: a marathon at the end of winter and then back to trail.

Ultra-Trail di Corsica, my first 100M race

So, as you know, I really enjoy trail running. Over the years, I’ve done quite a few fast-packing hikes, like the GR20, the GR54, and others, you can read about those adventures elsewhere on my blog. But last year, in 2024, something changed: I decided to go around Mont Blanc. Everything started at that moment, because I began my Mont Blanc tour the very day after the UTMB finished.

When I arrived, I was swept up in the excitement surrounding this legendary trail running race. The atmosphere was absolutely incredible, and it really drew me in. There was this amazing connection between all the runners, the organizers, and everyone involved. That’s when I thought, “This is really my year, I need to get into this.” I really wanted to do it. I felt it would be a fascinating experience, maybe similar to the fast-packing I love, but with a much stronger community element.

So, I completed my Tour du Mont Blanc in four days, and it was incredible. Afterward, I told myself, “Okay, now I’m signing up. I ultimately want to do the UTMB, a 100-mile format race.” About a month later, I went on the UTMB website and found a race that I liked…

That’s how it all started: I decided to sign up for the Ultra Trail di Corsica by Restonica Trail, a race of 110 km with 7,000 meters of elevation gain, right in Corsica. I realized that to participate in the UTMB, you need to collect “stones,” and that by racing in events on the UTMB circuit, I could accumulate the necessary stones to eventually get my spot. That’s how my journey began. I registered first for the “100 Miles” format, and for the preparation for this race, I had about 8 or 9 months. For context, at that time, I had just set a new PR on my half-marathon: 1:45.

I started with marathon training using the Runna app, running five times a week. After four months, I ran my marathon. Then, from March to July, I continued with Runna but switched to their 100 km training plan, also for four months. During this phase, I was running six times a week in Elite Plus mode on Runna, reaching a maximum of around 120 km per week. On top of that, I was doing CrossFit year-round, between three and five times a week. So, that was my preparation.

After that, I set out for the race. I left at the end of June, flew to Bastia, and stayed at a house there. The race started on Thursday evening, and I think I left around 3pm I had packed my bag with all the mandatory equipment. I went to collect my bib and my drop bag. In my drop bag, I put a change of clothes and shoes. Then, I just went and had a burger at a restaurant before heading to the start line.

The race started at 11pm in Corte. I was in the start corral, among 450 runners, positioned about three-quarters of the way back because I arrived just before the start. When the race began, it was absolutely incredible: many people lined the streets of Corte leading to the first mountain, lighting flares like at a football stadium. We were surrounded by red lights, lots of smoke, it was such an intense atmosphere. And so, I set off, with my pack, ready to go.

Corte, Thursday 23:00

And there we go, the race kicks off, and we immediately tackle the first big climb up to Padule, a solid 1000 meters of elevation right from the start. Since it’s nighttime, everyone switches on their headlamps, and soon we’re all strung out in a line, following each other up narrow single tracks. Overtaking is nearly impossible, so you’re forced to go at the pace of the group, which is both frustrating and kind of special. Not being able to set your own rhythm is tough, and it takes about two hours to reach the top.

At this point, everyone is still full of energy, myself included. I’m thinking, “Let’s go!” and soaking in the atmosphere. It’s my first time ever starting a race at night, and seeing the endless line of headlamps snaking up the mountain is just magical. We’re all climbing together, a moving constellation in the darkness.

At one point, I notice my bib number, 464, and spot someone ahead of me with bib number 87. I know that bib numbers are assigned based on the UTMB Index, which basically ranks your speed and experience as a trail runner. So I figure, “Okay, this guy is number 87, he must be pretty highly ranked. I’ll try to follow him and tuck in behind.” It’s a small strategy, but in these early hours, it helps to have a target and a bit of motivation.

Padule, Friday 00:57, 7km 1316m+ 1:57 218th

We finally arrive at Padule, the first aid station. I’m still sticking to my nutrition plan: at every aid station, I take two flasks filled with isotonic powder for a sugar boost, and I keep up with my 45-gram carbohydrate gels, one every hour. At Padule, I grab a quick Coke, don’t hang around, and get right back to climbing.

At this point, the pack starts to spread out. There’s a short climb left before the trail opens up onto a kind of plateau. Now, for the first time, I can actually start overtaking people and move at my own pace, since the trails are getting wider and a lot of runners are taking a break at the aid station. This is when I really start to get into the race, I’m feeling good, finding my rhythm, and overtaking other runners as I go.

Boniacce, Friday 01:59, 15km 1704m+ 2:59 186th

By the time I reach Boniacce, I’m really in the race, I’ve moved up to 186th place, already gaining quite a few positions. I’m still feeling great, running strong, and even clocking some kilometers at 12 km/h on the flatter, easier sections. It’s a huge confidence boost to be moving up the ranks and feeling so good this early on.

After Boniacce, the trail heads downhill, and I take advantage of it, descending quickly and overtaking a lot of people. There’s a short climb, then another descent toward Calacuccia. Calacuccia sits next to a beautiful lake, and as I approach, the sun is just starting to rise, around 5 in the morning. The sky turns red in the distance, with reflections dancing on the lake. It’s a truly magnificent moment, and I take a second to soak it all in.

Calacuccia, Friday 06:14, 33km 2366m+ 6:14 167th

When I get to Calacuccia, there are some slightly inclined sections, but I’m still in great shape and able to jog them. Up to this point, I’ve been power hiking the climbs at a solid pace, running the flat sections at about 10 km/h, and jogging the rolling terrain. I really feel at the top of my form, moving at a pace I’m proud of. At the aid station, I stick to my routine: refill my isotonic drinks, grab an energy gel, drink some Coke, and then head out again.

At Calacuccia, I’m in 167th place. From here, we start the biggest climb of the race, a single ascent of 1,700 meters of elevation gain. We pass close to Monte Cintu, the highest mountain in Corsica, and by the Cintu lake. It’s an incredible section! For this climb, I take my time and walk, as the sun rises and the temperature starts to climb, so I put on sunscreen.

It’s at this point that things start to get tough, and I realize I’m about a third of the way through the course, around 40 kilometers out of 110, after 7 hours of running. I think to myself, “I’m on track; if I keep this pace, I could finish in 21 hours.” That’s beyond my wildest dreams, my ultimate goal was 24 hours, so even if I can’t maintain this pace, maybe I can still hit that 24-hour mark. Based on my UTMB index, the estimate was around 26h30, so that was my realistic target, maybe trying to do a bit better to improve my index. Of course, my main objective is just to finish, but realizing I’m on pace for 21 hours feels incredible. I really feel like my training has paid off. I’ve never attempted anything like this before, and now, nine months after starting my journey, I’m in the middle of a 100-mile ultra.

The climb keeps getting tougher, the elevation is brutal. My previous max was a race with 3,000 meters of elevation gain, and now, at the top, I’ve already hit 4,000 meters, so I’m pushing past my limits. Along the way, I meet a Belgian runner. We chat in English (he’s Dutch-speaking), and he tells me he’s done twenty 100+ kilometer races, including the UTMB, and he’s just here to test himself. After a while, he says, “You’re too fast for me, I’ll let you go on ahead.” I don’t feel fast at all, climbing is really my weakness. I’m a relatively heavy runner (1.87 meters tall, about 87 kg), so the climbs are never my strong suit.

Eventually, we reach Cintu lake, which is about halfway or two-thirds up the climb, and then there’s a steep final ascent to Bocca Crucette at the spur of the scree. It’s very steep and long, but at the top I take a moment to enjoy the view. At this point, we join the GR20 trail, right next to the Tighjettu refuge, and we’ll follow the GR20 for three stages: Tighjettu, Ciottulu di i Mori, Manganu, and half of Manganu to Petra Piana. Now there are more hikers on the trail, and you can tell they’re all pretty impressed by what we’re doing.

Bocca Crucette, Friday 08:58, 44km 4154m+ 9:58 159th

From here, there’s a long descent down to the next aid station at Ballone. I start to notice that going downhill is getting a bit tougher. I keep jogging, but not as fast as before, at first, the descent is super steep, so I take my time. Some runners begin to pass me, really flying down the trail, and that’s when I realize my body isn’t as fresh as it was at the start. My legs are getting heavier; sometimes I want to step quickly, one foot after the other, but my legs just don’t respond as fast as I’d like. The muscle fatigue means I can’t always put my foot exactly where I want, which throws off my balance a bit, so I know I need to pay more attention.

At this point, I’m wearing my S/Lab Ultra Genesis shoes, and since I have a hallux valgus (basically, wide feet), they’re starting to feel pretty tight. I want to tighten my shoes to keep my feet stable on the downhill, but at the same time, the pressure is getting uncomfortable. I start to feel some real discomfort, but I try to go over it, telling myself it’s just a minor annoyance and not a big deal. Still, it’s the first sign that things are starting to get harder.

Ballone, Friday 10:18, 48km 4154m+ 11:18 155th

When I arrive at Ballone, it’s a refreshment point, and I’m now in 155th place. I put on some more sunscreen and grab a drink.

I remember a volunteer at the aid station brought me a drink, and as I thanked her, she asked, “Where is your bib?” I actually had my bib on a holder at my side, but she asked me to move it to the front. I was a bit reluctant, when the bib is in front, it gets in the way on climbs, pressing against your stomach and thighs as you bend forward. Plus, it can damage the bib, and I like to keep mine in good condition to display at home afterward.

She pointed out, “Yes, but you could get a penalty. Besides, they already made me wear the t-shirt and everything.” That’s when I realized she was a race official, with the authority to enforce the rules. According to the official regulations, your bib must be visible in front. So, I moved my bib to the front, but honestly, I was already starting to disconnect a bit from the race and the time, especially as I stood there being reminded of the rules. All I wanted was to be left alone and keep moving, it wasn’t easy.

Still, I’m at Ballone, 50 kilometers into the race and still gaining positions. From here, we head into a climb up to Ciottulu di i Mori, the GR20’s refuge. I’m feeling a bit tired, so I go slowly. We’re climbing, and eventually reach a pass.

At the top, I think, “Yes, OK, I made it.” But when I look to the right, I see the route keeps rising along the mountain ridge. I remember this really gave me false hope, I thought the uphill was over, but it wasn’t at all.

Ciottulu di i mori, Friday 12:44, 53km 4860m+ 13:44 176th

Eventually, I arrive at Ciottulu di i mori. There’s only water at this aid station, so I just refill my isotonic drinks and keep going. By now, I’m in 176th place, I’ve lost about 20 spots, and it’s really starting to get tough. Next comes a descent to Ciattarinu, the life base where we get our drop bags. I know these sections well because I’ve already done the GR20, but now I’m moving really slowly. I can’t run anymore, even though the terrain isn’t technical, just a bit rocky and unstable, but not steep. It should be runnable, but I just can’t do it. My feet hurt, my legs feel heavy, and people are passing me.

At this point, I realize my 24-hour, 26-hour, or similar goals are gone. I had hoped to get to Ciattarinu around noon on Friday, but I only make it at 2:30pm, after about 15 and a half hours of running, when I thought I’d be there after 12 hours. So when I finally reach Ciattarinu, I’m pretty tired and fully aware that I won’t achieve my target times. But since it’s the life base, I decide to take some time to recover.

Ciattarinu, Friday 14:30, 60km 4905m+ 15:30 189th

Finally getting there, it really hits me: from this point on, my only goal is just to finish the race, no matter what happens. I’m already walking, and if I’m walking at this stage, I know I probably won’t be able to run much for the rest of the course. I’m at 60 km out of 110, with 5,000 meters of elevation gain done, so I know I have about 2,000 meters of ascent and 50 km left, which means I’ve already covered a significant chunk of both the distance and the climbing.

Ciattarinu is a proper life base, so you can take a nap, shower, and reset. I head into the changing room and take a shower (I didn’t have a towel, so I asked someone to lend me one). Changing clothes helps me feel a bit better. Then I’m faced with a dilemma about my shoes: my Salomons are hurting my feet, but the Hokas I packed are a bit worn out. Still, the pain from the Salomons is too much, so I switch to the Hokas Speedgoat 5, hoping for some relief. While at the base, I also eat a bit, there are more substantial food options, so I grab some pasta salad, iced tea, and soup.

I take a break, and just then, a real storm hits, pouring rain, thunder, and lightning. While we’re waiting it out, I chat with other runners. People are discussing how much distance is left, the storm, and whether or not to rest, since there are beds available. There’s a real possibility to take a proper break before heading out again. At this point, I realize I’ve been awake for about 30 hours straight, woke up Thursday at 10am, and now it’s Friday, 4pm. But I tell myself, no, the goal is to keep moving; I don’t really feel like sleeping.

I wait for the storm to pass, and when the rain eases, I set off again. I join a small group, and although the storm is still around, it’s far enough away, I can see the lightning and count the seconds between the flash and thunder, so I know it’s not right on top of us. I put on my rain jacket for protection, and actually, I feel pretty good, able to jog the flat sections as we head toward Lake Ninu.

Then, something happens that makes the rest of the race really tough: I start feeling pain on the inside of my right knee. I think it’s runner’s knee, like IT band syndrome, but later I learn that IT band syndrome is on the outside, and pain on the inside is actually due to the pes anserinus. That pain sticks with me for the rest of the race. I keep going, but I really can’t run much anymore, at best, I can jog at 7 to 8 km/h on the easiest flats, but that’s it. Even then, I feel exhausted, so I don’t push too hard.

After an uphill, It stops rainning and I’m now walking to Inzecche around Ninu’s Lake, it’s beautiful.

Inzecche, Friday 19:33, 74km 5518m+ 20:33 177th

When I arrive at Inzecche, everyone is super friendly. I joke around with the volunteers, they tell me, “Oh, you’re Belgian! I met a guy this morning who asked if I had some Chimay,” and so on. The atmosphere is great, and it’s a nice boost. It’s around 7:30pm, so there’s a little daylight left, but not much.

At this point, I’m no longer following my original nutrition strategy. I still refill my bottles with isotonic drink (about 30 grams of carbs per bottle), but I’ve stopped taking gels, since I know I’m not going to finish in the time I’d planned. I think, “Okay, it doesn’t matter, I’ll just keep walking.” At each aid station, I eat more solid food and drink some Coca-Cola. I’m enjoying the Naak waffles (like honey waffles or stroopwafels from Belgium), which I know I can eat without any problem. I even have a small tartine with wild boar terrine, amazing.

Then I set off again. It’s 8pm, and I witness my second sunset, heading out for a second consecutive sleepless night. By now, I’ve been awake for around 36 hours. I’m still on the plateau around Lake Ninu, which is relatively flat with some wet areas. There are lots of cows, some wild horses, it’s really beautiful. I keep walking, but I start feeling pain in my feet. When I take off my socks, I see they’re soaking wet, so I change them. (Back at Ciattarinu, when I changed socks, I was using Merino wool toe socks, but they were torn at the big toe, so I switched to standard Decathlon socks. Now, since my feet are wet and hurting, I change socks again.)

At this point, I tell myself, let’s go! I know what’s ahead: two climbs of 750 meters and 40 kilometers left to finish the race. I know the first 750-meter climb well, it’s part of the GR20, specifically the ascent from the Manganu refuge to the Capitellu breach, one of the toughest sections of the GR20. So I know what’s coming.

I keep walking, and at this stage, I’m really all alone. Since the rain stopped, there are hardly any people left on the course. I’m completely alone in this plain, making my way toward the Manganu refuge. An Austrian runner, Dominik, passes me and chats a bit, but I’m deep in my own head and not very talkative. I just want to keep moving forward and finish the race. I’m pretty convinced I’ll make it. My knee hurts, and my muscles are tired, but it’s not unbearable. The knee pain is there, but it’s not sharp or electric. I’ve had tendonitis before (like in my Achilles), where I couldn’t even walk, this is different, just a dull ache. So I keep moving, reach Manganu (without stopping), and by now, night has really fallen. I take out my headlamp and start the climb up to the Capitellu breach.

As I was saying, I know exactly what awaits me on this climb. It’s a section that feels truly endless: every time you believe you’ve reached the top, you realize there’s still another stretch left to go. The summit is just 750 meters away, at an altitude of 2,000 meters, but the challenge isn’t the distance, it’s the terrain. The path is extremely technical, made up entirely of rocks and loose stones that require constant scrambling. Sometimes, the steps are so big it feels like climbing three stairs at once. I’m fully aware this is what lies ahead.

I have my music playing quietly on my phone speaker, just enough for me to hear, no one else. Night is falling fast, and as I climb, I spot other headlamps sparkling both ahead of and behind me. There’s something reassuring about that sight; it’s a reminder that I’m not alone in taking on this punishing segment. Climbing this at night is an entirely different experience, intense and impressive in equal measure.

Fortunately, the route is well marked with small retro-reflective markers, which makes a huge difference. I hadn’t realized before that some official GR20 trail markers are also reflective, that really helped guide my way, and I had no trouble staying on course. Still, the climb seems even longer at night, and every meter feels like ten.

Not long after, another runner appears, a guy who’s a bit taller than me. He starts following my pace closely, practically stepping where I step, “taking my wheel” just like in cycling. We don’t really talk much, he answers sparsely, only breaking the silence occasionally. Every now and then, I’ll announce, “Dude, I need a water break,” and we stop for a quick pause. The climb is so steep and demanding that every four or five steps I find myself grunting like a tennis player, which I imagine is pretty funny for anyone following.

Eventually, as we near the top, I ask, “Where are you from?” and he replies, “Spain.” Since I can speak a little Spanish, I switch languages, and we start chatting. His name’s Xavi (Javier), and we share a bit of advice with each other, “Cuidado en este tramo,” “Be careful here.” The summit of the ridge finally comes into view. Even though it’s night and I can’t see much, I know from memory that this spot overlooks the stunning twin mountain lakes, Lake Capitellu and Lake Melu (see below what it looked when I hiked the GR20 in 2020).

From here, though, the challenge isn’t over. Now comes the technical ridge traverse, almost like mini-climbing, but this time heading downwards. The descent is so steep that we’re forced to sit, slide, turn around to find footing, it’s an adventure in itself. A bit later, Dominik, the Austrian, catches up and joins us, so now we’re a trio tackling this wild stretch together. We can’t help but curse and laugh at the absurd difficulty of this section, “What is this madness?” someone exclaims, and it lightens the mood.

Ahead lies Bocca a Soglia, a mountain col a little further on. From there, we’ll leave the GR20 and begin the descent toward Lake Melu, while the official GR20 route continues down toward Petra Piana. The toughest stretch of the night is behind us, but the adventure’s far from over.

Bocca a Soglia, Saturday 00:06, 85km 6233m+ 25:06 177th

We finally arrive at the aid station, situated on the pass. It’s a minimal setup, just a liquid aid station offering Coke and water. As usual, I grab a little Coke to perk myself up and take a moment to refill my water bottles. Xavi is already there, clearly still full of energy. “Let’s go, man!” he urges, eager to hit the trail again. I quickly apologize, “Sorry, just need to top up my bottles,” and he patiently waits for me. As soon as I’m ready, we begin our descent together.

Right from the start, Xavi sets a fast pace for the way down. I do my best to stick with him and stay close, but after a short while, he outpaces me and gradually disappears ahead. The descent to Grotelle is short but intense: only about 3 kilometers, but with a dramatic 600-meter drop in elevation.

At this point, Dominik catches up, and he and I descend together while Xavi moves ahead at lightning speed. Xavi mentioned something about a friend waiting for him at the bottom, I’m not sure I understood everything, but it seems like he has an extra bit of motivation spurring him on.

The trail itself is steep enough that ladders have been fixed into the rock in certain spots. Navigating these ladders in the stillness of night, probably around 2 or 3 in the morning, adds a thrilling twist to the adventure. Despite the nerves, we make it safely down the ladders and continue on.

Then, mid-descent, I suddenly slip, the kind of moment where everything happens in a heartbeat. I just manage to catch myself with my trekking pole, but I hear a sharp snap: my trusted carbon Leki pole has broken. Now I’m down to a single pole for the rest of the race. I’m somewhere between three-quarters and four-fifths of the way through, so at least the finish isn’t too far off, but I know how much that pole has been helping with balance and stability. It’s definitely frustrating, but I remind myself this kind of thing happens in ultrarunning, you just have to adapt and keep pushing forward.

Grotelle, Saturday 01:29, 89km 6241m+ 26:30 176th

I press on and finally arrive at the Grotelle aid station, together with Dominik. Both of us are feeling surprisingly good, considering the distance we’ve covered so far. Dominik immediately says, “Yeah, let’s keep going,” and I agree without hesitation. “Yes, man, let’s continue. We won’t take a break, and there’s no chance we’re sleeping, but I do need to eat a little something as we move.” We quickly mingle with a few others and exchange some words with the volunteers, who, as always, are incredibly friendly and encouraging. Their positive energy is a welcome boost.

We know at this point that we’ve reached the very last climb of the race. To put it in perspective, we’re at kilometer 90 out of 110, just 20 kilometers to go, with one final ascent of about 750 meters still ahead. I remind myself of some training I did earlier in the week back in Corsica: I managed 750 meters of elevation gain in an hour and a half, round trip, not far from home. I try to channel that experience, thinking, “All right, it’s time to finish this off.”

So, off we go again, and soon we’re joined by another runner, his name escapes me, but he slots right into our little group as we start the climb. Interestingly, the section begins with a short descent before the real uphill challenge begins. It’s probably around 2 or 3 in the morning, and out of nowhere, we spot Xavi, the Spaniard, heading towards us in the opposite direction. He stops to ask, “Ah, ¿has visto a mis amigos?”, have you seen my friends? I’m thrown off, replying, “Man, what are you talking about?” He seems confused, unsure which way to go. I watch as he turns around and doubles back, he must be three kilometers into this segment already, but now he’s going back, meaning he’ll cover an extra six kilometers just searching for his friends. I can’t help but think, “His friends are definitely on the course and will find him at the finish line. Not sure what’s going on with him.”

We carry on, and once again I see little headlamps bobbing in the far distance above us. The climb looks endless, and I convince myself, “No way do we have to go all the way up there.” But the ascent turns out to be just as brutal as it looks, an unrelenting effort. Dominik takes over leading, and I follow in his footsteps, our energy running low. I’m so exhausted that even taking a deep breath makes my chest ache strangely, like I can’t get a full breath in. Still, the only option is to keep moving forward, even as the trail seems to go on forever.

So there I am, relying on my single remaining trekking pole as I climb. I use my left arm to help, pushing off rocks whenever I can, just focusing on making steady progress. We keep inching forward, each step taking more effort than the last. Eventually, we reach a section that’s noticeably steeper. I think, “This has to be it, the final pitch.” It’s so steep that each step feels like climbing four stairs at once.

Out of nowhere, I start to overheat. An intense, overwhelming heat floods over me, making me consider stripping off my T-shirt just to cool down. But then, I remember those stories from Everest movies, where people with altitude sickness start feeling so hot they take off all their layers, even when it’s freezing, and it never ends well. That thought pulls me back, and I decide against it. Instead, I just keep pushing on, but a little alarm bell goes off in my head: “Can I really trust my body at this point?” The fatigue is catching up, after all, this is my second night without a wink of sleep and I’ve already covered 100 kilometers. Obviously, I’m exhausted, but I keep climbing regardless.

The overheating continues all the way to the summit, but finally, we make it. At the top, we find ourselves on a plateau, the Alzu plateau. From up here, we spot the aid station in the distance, but even now there’s still about a kilometer of trail before we reach it. We’re at about 1,700 meters of altitude, so the air isn’t as hot. Instead, the wind picks up, and as soon as we stop climbing, my heart rate drops, and suddenly everything shifts: what was a burning heat is replaced by shivering cold. I start to shake, my teeth chattering uncontrollably, I’m freezing. I ask the guys with me, “Are you cold?” but they shake their heads: “No, we’re fine.” I’m surprised, but at least the aid station is visible ahead.

I focus on reaching it, but the cold just keeps creeping in. I stop for a moment, turn to my friends, and confess, “Guys, I’m really cold.” I pull out my rain jacket, still damp from earlier in the race, when it had rained, but I throw it on anyway, hoping it’ll help trap some heat. Gradually, as I keep moving, I start to warm up, and I press on, determined to reach the comfort of the aid station.

Alzu, Saturday 05:03, 99km 6969m+ 30:03 167th

When I reach the Alzu aid station, I immediately find a seat in one of the camping chairs and wrap myself up in two blankets that the volunteers provide. Even bundled up, though, I can’t stop shivering. The chill seems to go all the way to my bones. At that moment, a wave of doubt washes over me: I’ve already covered 100 kilometers, just 10 kilometers stand between me and the finish, but my body suddenly feels like it might give out. I seriously wonder if I’m about to have to drop out of the race, right at the end.

I recall the minor muscle pains I’ve had throughout the night, nothing too severe, but still, not exactly ideal either. The thought crosses my mind: I don’t want to risk putting myself in danger, and if the shivering doesn’t stop, I might have no choice but to stop here. If it means staying safe, I’ll do it.

As I sit there, the sky begins to lighten, and I see the first rays of sunrise. It’s 5am I turn to Dominik and the other guy who made the climb with us and say honestly, “Listen, guys, I can’t carry on right now. Go ahead without me. I’m going to wait here until the sun rises a bit more, and it warms up. Maybe I’ll rest for an hour, but it doesn’t make sense to set out again while I’m still shivering.” They nod, wish me luck, and move on without me.

Left alone, I check my phone and see a message from the race organizers on WhatsApp: “Ah, Mr. Picard, where are you? We can’t find you, etc. Did you drop out? Please inform us as soon as possible.” I’m puzzled, what’s going on? It turns out my bib hasn’t registered at any checkpoints since Inzecche, which is before Manganu and before the climb up to Breach of Capitellu. It’s been hours since my last chip scan, so as far as the organizers know, I could have dropped out way back at Inzecche, around 7:30pm.

I explain, “But I’m right here.” One of the volunteers catches wind of the situation, grabs his radio, and reports, “Yes, bib 464 is here.” My bib is attached to my belt, so I hand it to him for confirmation. The conversation gets surreal: over the radio, they insist they haven’t seen me since Inzecche. Someone says, “He dropped out.” The volunteer protests, “No, he’s here in front of me. Bib 464, present and accounted for. He hasn’t dropped out.” More chatter with the Corsican accent follows, “Tell him to take out his soft flasks; sometimes the bib gets wet and stops working.” The volunteer confirms, “I have the bib in my hand, there are no flasks attached.” We all laugh a little at the situation. Eventually, the most logical explanation is that the rain after Ciattarinu affected my chip, stopping it from registering correctly. At least the confusion gets sorted out in the end.

Meanwhile, I’m still cold, sipping iced tea and relaxing under the blankets, debating my next move. Do I try to catch a little sleep? But with just 10 kilometers left, I worry I might not wake up in time to finish. The cutoff is 11pm, and it’s only 5am, but I don’t want to risk missing the end. I remind myself, “I am not going to give up. At worst, I’ll take a long break and then set out again.” But in my heart, I know I want to finish as soon as I can, not just rest here on the edge of the finish line.

At that point, I found myself sitting at the aid station, still shivering, in 167th place. I had been on the move for 30 hours, completed 100 kilometers, and climbed over 7,000 meters of elevation gain. Dominik had already pressed on ahead. While I sat bundled up, who should appear but Xavi, and this time, he wasn’t alone. With him were his two friends. As they arrive, I quickly realize his friends aren’t race participants; they’re Spanish as well but living in Corsica, and they’d come up to meet Xavi at Grotelle, so they could join him for the last two stages. Suddenly, it all made sense, no wonder Xavi had doubled back earlier on the trail to find his buddies.

The three reach the aid station and both of his friends, who speak French, quickly strike up a conversation with me. One of them, noticing I’m still cold, offers me a really warm jacket. I gratefully pull it on and soon feel the warmth returning. It’s a fantastic gesture, and being reunited with Xavi, my buddy from earlier, definitely lifts my spirits. Re-energized and with my new crew, I decide it’s time to set off again.

Around this same time, I’m relieved that the bib situation finally got sorted out. I’d been a little worried because I knew my family was following my progress, especially since I’d told them I’d finish in 30 hours at most, and they hadn’t seen an update for quite a while. Now that I had a signal, I shot off a quick message on WhatsApp, reassuring them that everything was fine, and I was back on the move.

We set off into what became my second beautiful sunrise of the race, just as we began descending from Alzu. The descent to Corte starts off steep but then gentles out, stretching for 10 kilometers and shedding 1,300 meters of elevation through serene woods. I kept pace with Xavi and his two friends, who were absolutely flying down the trail. Somehow, I managed to stick with them, even though I hadn’t really run much in the last 60 kilometers, being swept up by their energy and the excitement of the finish being so close.

We sped downhill, making great progress, and at one point I even overtook Dominik thanks to our group’s running pace. I pushed myself to jog as often and as fast as I could, repeating, “Come on, just a few kilometers left,” the adrenaline fuelling every step toward that long-awaited finish line.

At that point, I start experiencing what I’d call visual misalignments, a sort of distorted perception. I wouldn’t go so far as to call them hallucinations, but my mind definitely starts playing tricks on me. For example, I might spot a rock ahead and my exhausted brain immediately insists it’s a cabin. Deep down, I know it’s not a cabin, I’m aware of my extreme fatigue, but that automatic part of my mind is convinced. To prove to myself it’s just a rock, I need to zero in on the object, analyze its details, and consciously override my tired mind. This kind of mental effort, bringing reality back into focus, quickly becomes exhausting in its own way.

By now, this is happening constantly: shapes in the trees start to look like animals, ordinary tree trunks seem to morph into strange creatures. Even though I rationally know these things aren’t there, the experience is still weirdly disorienting, and it takes real focus to ignore the illusions. The constant battle to convince my brain of what’s real is draining as I approach the final kilometers. Xavi and his friend are long gone, I couldn’t keep up with them.

I try to keep jogging, but my pace is painfully slow. Dominik, who had caught up to me again, breezes past, saying, “I’m hurting everywhere, but I just want to finish, I want to get home, I want this to be over, so I’ll run no matter what.” Inspired, I try to emulate him, pushing myself to run and overcome the pain, but my body just isn’t having it.

Eventually, we reach an area very close to Corte, a well-known spot for hikers, part of the Northern Mare a Mare. Now the trail is bustling with local trail runners, out for their usual Saturday morning runs. They cheer us on: “Come on, you’re almost there, only 2 or 3 kilometers left!” I greet them with a thumbs up, but in my head, I’m thinking, “Guys, stop telling me it’s just 2 kilometers, I know that’s not true, and honestly, even those last 2 kilometers feel impossibly long right now.” Nonetheless, all I can do is keep moving forward, because at this stage, that’s all that matters, just keep going, step after step.

In any case, the encouragement from everyone I pass really lifts my spirits, all along these last few kilometers, people call out words of support, and it genuinely warms my heart. Despite that, the walk to the finish seems to drag on forever. The trail undulates with small climbs and dips, nothing huge, just about 160 meters of elevation gain scattered across the final ten kilometers, but after so much distance, every little ascent feels like a mountain and every descent tests tired legs.

As I draw closer to the finish, I still can’t quite believe it, I’m actually going to make it. We hit the outskirts of town; volunteers line the route, clapping, and there are people sitting at cafés who cheer us on, making the atmosphere feel celebratory. The path becomes easier, and with that surge of energy (maybe just pure adrenaline), I find myself starting to run again. I’m running through the streets, feeling a new wave of energy and excitement.

At last, I see the finish line arch coming into view. The end is right there, so I dig deep and pick up the pace, maybe I’m sprinting at 12 km/h, maybe it’s less, the details all blur together. I cross under the arch, hearing the announcer call out my name, and as I pass through, the feeling is pure, overwhelming relief. I can’t help but scream with joy, I’m just so happy. I did it! In 34 hours, well, 33 hours and 52 minutes, I finish in 173rd place. The sense of accomplishment is incredible. It was truly an amazing experience.

Corte, Saturday 08:52, 112km 7128m+ 33:52 173rd

As soon as I finish, I’m reminded of what I was told back at Alzu, that my race bib chip was broken. Still, I collect my finisher’s medal, my race t-shirt, and soak up the moment. After that, I head straight to the race organizers to explain: “Look, my bib chip isn’t working.” They jot down my finish manually, taking note of my time and details, which means my result is official, just not tracked by the chip. The only downside? I don’t get the classic finish video, since the time didn’t register exactly as I crossed the line, it all depends on the chip. But honestly, after this adventure, that seems like such a small detail.

At the finish area, I spot Dominik sipping a Coke, Xavi celebrating with his friends, it’s amazing to see everyone there, sharing in the achievement. The whole experience feels truly exceptional. This was my very first 100-mile race, and it delivered a real reality check. I’d gone in thinking I’d finish in 26 or 27 hours, 30 at the very most, but I ended up clocking 34 hours, much of it walking. Regardless, it was an incredible adventure, and I know I’ll be back for more, aiming to run a bit more next time.

Few days later at the airport, as if to put it all in perspective, I bump into Lars, the Belgian runner. He mentions he’s finished twenty different 100-kilometer races, and has even completed the UTMB before. His take on this event is reassuring, he tells me this race truly stands out, that it was more difficult than most because of the relentless and technical terrain. “Extremely tough and complicated to run,” he says, “especially for a first 100-miler.” Hearing that, I realize just how proud I should be for finishing such a challenging race…

See you in Slovenia for my next 100M race!

Vibe coding the migration of this blog from Express to Next.js

If you’ve been following my articles, you’ll know that I created this website three years ago as a way to learn JavaScript programming. Back then, I followed an Express.js tutorial and applied it to build this website. While functional, Express.js is relatively outdated compared to modern frameworks that make developers’ lives easier. I experienced this firsthand when I built Urbanex using Next.js and React, the development experience was much more pleasant. Next.js’s built-in routing, server-side rendering, and component-based architecture made development smoother and more intuitive. Additionally, Next.js offers better navigation and performance out of the box, without requiring extra configuration or third-party libraries. However, the main reason I decided to migrate this blog to Next.js was to host it for free on Vercel, compared to the $7/month I was paying on Heroku for the Express version.

LLM-assisted coding: a game changer

When I initially created this website, we were in the GPT-3 era, and LLM-assisted coding wasn’t mainstream, at least not for me and my peers. Today, it’s a different story, I use LLMs daily through Cursor for coding, and it’s transformed how I approach development tasks. My first step in migrating the website was to open the prompt dialog and type:

I want you to convert this whole project into next js + tailwind

With its latest update, Cursor now offers an ‘agentic’ way of interacting with the LLM, automatically analyzing the codebase and adding relevant project files to its context. This is a significant improvement over manually copy-pasting code, making LLM-assisted coding (or vibe coding?) much less cumbersome. The LLM created a Next.js project with basic routes and a boilerplate Tailwind integration. Here’s its first implementation of the blog page, which serves as the home page of this website:

import React from "react";
import Link from "next/link";

export default function BlogPage() {
  return (
    <main className="min-h-screen p-8 max-w-4xl mx-auto">
      <h1 className="text-4xl font-bold mb-8">Blog</h1>
      <div className="space-y-4">
        {/* Blog posts will be listed here */}
        <p className="text-gray-600">Coming soon...</p>
      </div>
    </main>
  );
}

Thus, I had to follow up with (yes, with the typos):

now find a way to include all my blog posts, they are MD, display them, also each blog post as a title that should be listed on the index

The LLM suggested installing npm install gray-matter react-markdown and wrote helper functions to parse my blog posts written in static markdown files.

While coding, Cursor’s agent would often fix errors automatically thanks to its linter integration. The agent proactively fetches and addresses issues, which is much more pleasant than manually running the code, encountering errors, and having to copy-paste them for debugging.

As I ran the website locally, I enjoyed a live preview of the results as the agent coded. This was particularly satisfying since I typically work in data/analytics/cloud engineering rather than frontend development, where I don’t get such immediate feedback. With web development, the LLM’s impact is very tangible.

The limitations and lessons learned

While I’m more efficient at coding with LLM assistance, I’m grateful that I can understand what it’s doing to perform quality checks. For instance, I encountered a styling issue and noticed that the LLM had used an outdated version of Tailwind. I manually updated it to use the latest version. Since Tailwind 4.0 was only released earlier this year, with just four months of existence, LLMs don’t have many examples of its implementation in their training dataset.

Additionally, the LLM would often fix and code things with local view only. As an engineer, I could sense when it was going down a rabbit hole, focusing on symptom-fixing or avoiding necessary abstraction/modularity. For example, I used this prompt:

analyse the page and see how they all have some kind of title in a first div with some styling (lighter bg) then a main content.
shoud this be refactored so that this styling is defined once? using a comopnent or so?

Leveraging my experience with Next.js, I could guide the LLM in the right direction, and its output was satisfactory. While LLMs are powerful tools, they work best when paired with human expertise and intuition.

My experience with LLM-assisted coding extends beyond this migration, and I can confidently say that it’s not equally effective with all technologies. For example, it struggled significantly with Dockerfile creation, making numerous errors. This highlights an important point: LLMs excel at tasks with clear patterns and abundant examples in their training data, but they can struggle with more complex or niche technical challenges.

So, can anyone vibe code their next startup?

If it’s a simple Next.js web app, maybe. But here’s what I’ve learned from this experience:

1. Know your tools: Understanding the underlying technology is crucial. LLMs can help you code faster, but they can’t replace fundamental knowledge.
2. Quality control: Always review and understand the code generated by LLMs. They can make mistakes or suggest suboptimal solutions.
3. Iterative development: Use LLMs as part of an iterative process. Start with a basic implementation, then refine and improve it.
4. Domain expertise: The more you know about a technology, the better you can guide the LLM and evaluate its suggestions.

This aligns with my overall impression of LLMs’ usefulness: it can do what I already know how to do, and makes me faster and more efficiently with LLMs. If there’s a mistake, I can fix it because I understand the underlying concepts. I don’t trust LLMs blindly to the point where I’d be comfortable with results I don’t understand due to a lack of skills. Maybe one day, but for now, my programming knowledge is what makes LLMs truly valuable to me.

Overall, I’m feeling lucky to have both learned and practiced programming before the era of LLMs. This traditional path provided me with a deep understanding of programming concepts, debugging techniques, and system architecture that I rely on daily. Years of hands-on experience have given me the intuition to spot potential issues and the ability to think through complex problems from first principles. This foundation allows me to critically evaluate LLM-generated code, identify potential issues, and understand the “why” behind certain implementations. Simultaneously, I’m fortunate to be working during this technological transition, where I can leverage LLMs to accelerate my workflow while still applying my expertise. This combination of traditional programming knowledge and modern AI tools makes me more effective in my work.

I also see great value in improving LLM-assisted UX, as Cursor is doing. I can’t wait for a good email client with an integrated LLM agent, that would handle meeting planning or draft basic answers for chores related emails!

Anderlecht is the slowest while Boitsfort is the fastest, permit delivery wise

During last spring, I had the incredible opportunity to attend the London based Startup incubator, Entrepreneur First. The experience was quite extraordinary, allowing me to meet many brilliant minds, and also amplifying my understanding of innovation. For anyone curious to sample the atmosphere of this unique entrepreneurial environment, I recommend the article – Inside Entrepreneur First: a survival guide for the world’s most exclusive talent accelerator.

After spending three months in the incubator, I found myself on an unexpected path. Despite my exhaustive attempts, I couldn’t match with the right co-founder to share a common vision. Hence, I made the choice to return to Brussels. Following intensive customer development, brainstorming and idea-discussions, I wanted to feel pragmatic productivity and build something tangible.

I directed my energy to a subject that had intrigued me for a long time: the Brussels’ real estate landscape, primarily due to my involvement in multiple related projects. Dealing with property in Brussels introduces one to the intricacies of urban planning. If you were to renovate a building, you’d have to secure an urbanism permit by presenting the project to the respective municipality for approval. It is fundamental to note that construction cannot commence without possessing this permit.

To facilitate this process, Brussels launched a permit exploration portal, OpenPermits.brussels, powered by a public API. But it fell short of offering some elements I was searching for. That’s how Urbanex was born – to supplement the public permit portal with some enhance features, thereby enriching the user’s interaction.

An overview of permits’ lead times

The process of permit application involves a specific timeframe within which the managing authority should convey their decision – referred to as lead time. Having recently applied for a renovation permit myself, I was interested in finding out if these time limits were adhered to.

A helpful metric to gauge this is the ‘lead time deviation’ – the difference between the expected and actual duration. The graph below shows the permits’ yearly rolling average lead time deviation.

What a journey it has been! Brussels used to have almost a year of delays but is now delivering permits even before their deadline. The total volume of permits submitted has remained quite stable, except during the COVID period. However, it looks like a step change happened at the end of 2020 as seen in the graph. While the municipalities have probably improved their processes, another way to avoid missing deadlines is by extending them. Let’s explore their evolution.

While the main lead times were 45, 75, 90, and 120 days until 2019, they have increased to 75, 90, 160, and 190 days since 2021. The transition occurred in 2020, at the same time as the step change in the mean lead time deviation. The government likely extended the permit deadlines. Permits that used to be 45 days long now take 75 days, and those that were originally 120 days have been extended to 160 or 190 days.

To appreciate the efficiency of each municipality, I conducted a comparison of how many days prior to their deadline they managed to deliver the decision. Indeed, each of the 19 Brussels municipalities handles its own permits. For this analysis, we will focus on the data from the last year.

It’s impressive to see that all municipalities are delivering permits ahead of their deadlines. Boitsfort, the most efficient municipality, has an average pre-deadline decision period of 70 days, meaning that, on average, it will send its permit decision more than two months ahead of the deadline. However, it’s essential to consider that different permits have different expected lead times. Let’s break it down based on the planned time limits.

So, if you are introducing a 160 days permit in Saint-Gilles, you can expect your answer almost three months before the deadline. On the other hand, if it’s a 90 days permit in Uccle, you’re looking at 3 days of delay.

My personal experience: the inspiration behind Urbanex

So why did I build Urbanex, when OpenPermits is available? It’s all based on personal experience.

Urbanex is an online archive of urban planning permits

A few years ago, I bought a newly-built studio in Brussels. Upon moving in, I noticed that the facade lacked planned vegetation. To confirm what was supposed to be on the facade, I decided to refer back to the building’s plan. To locate these plans, I had to:

• Schedule an appointment with the municipality, which could take up to 2–3 weeks.
• Visit to the town hall to meet the official who had access to the paper archive of the permit, including its original designs.
• Check the archive, identify relevant document and scan them

The entire process cost me 10€ for the initial consultation and then an additional 15€ per scan. I ended up getting two plans, putting the total expense at 40€. This effort confirmed my suspicion - the facade was meant to be adorned with vegetation.

Although the traditional route eventually provided the information I needed, it was time-consuming and somewhat antiquated, especially given that permit documents are usually submitted electronically these days. Moreover, when a permit is first submitted, the associated documents go through a period of public consultation, allowing anyone to view them and voice their opinions.

However, once this review period is over, access to these documents becomes limited, with archives typically solely existing in paper form at the local municipality. Had I retained the required plans at the time my own flat underwent this consultation process, the taxing visit to my municipality would have been needless.

This made me realize that documents submitted electronically should remain easily accessible even after the public consultation period.

To fill this gap, Urbanex now ensures these documents are readily available, digitally archived, and can be conveniently accessed at any time, from any location, with no appointments. Navigating urban planning in Brussels just got a lot easier.

When AI meets urban planning

Then, I embarked into a journey to renovate my grandmother’s house, I intended to convert the attic into a liveable space, creating a duplex on the top floor. However, my architects cautioned me about the potential repercussions this kind of transformation could have. Such changes significantly increase the building volume and neighbourhood density, raising potential concerns from the municipality, especially if not executed thoughtfully. The dilemma I faced was whether to abandon my duplex dream or risk potential permit denial.

At that moment, I wished I could review past permits approved by the municipality, specifically for similar projects about raising the roof. With access to this information, I could have better understood the local authority’s stance on such transformations. Each permit application typically includes a brief description of its purpose. If there were a way to filter and search these descriptions, it would be ideal…

That’s exactly what I built.

By harnessing the power of Artificial Intelligence, I have made it possible for users to enter descriptions into Urbanex and filter results based on their particular needs. Let’s say you plan to construct a swimming pool in your backyard; with Urbanex, you can check how many similar projects have been accepted or rejected close to your place. This feature provides a better understanding of the feasibility of your project, potentially saving time and effort during your permit application process.

An ocean of opportunities

The journey of creating Urbanex has been a thrilling one. For me, it is again the opportunity to practice web development, using Next.js this time! Eventually, allowing me to create the interface between AI tools and the user. I have many more ideas about how else AI can bring value to this permit planning, I even aspire to, some day, craft a tool that can forecast permit approval outcomes.

My hope is that Urbanex proves as beneficial to you as it would have been for me. I appreciate all feedback.

So you want to Trail Run?

Over the last years, I picked up some experience in trail running. Or fastpacking. Or ultra-light trekking. Whatever you want to call it, I describe it as going on hike at a faster-than-normal pace. Indeed:

• In 2020, I hiked alone the GR20 in 6 days, Corsica, France
• In 2021, I completed the Alta Via 2 in 6 days, Dolomites, Italy
• In 2022, I solo-toured the GR54 in 5 days, Massif des écrins, France

Thus, in this post, I would like to share about my experience in this domain.

Do what you enjoy

It’s important to remember that trail running, or any multi-day hike, is meant to be enjoyable. Whether you’re a working professional or a student, this kind of trip is likely a choice, a tradeoff with other types of vacations such as a beach trip with friends or city tours. It’s a holiday. And if it is not, i.e. you’re a professional trail runner, then it’s unlikely that I have anything to teach you! Therefore, keep in mind that it is your trip and you should have it your way.

When I first attempted “trail running” in 2018, I planned to complete the GR20 reversed in seven days. I had a few weeks off in early September before starting my new job at McKinsey & Company, and I wanted to make the most of my time. I decided to spend two weeks in Corsica, enjoying the mountains and relaxing on the beach before working hard in top tier consulting.

I noticed that the GR20 stages were typically planned for 3 to 6 hours with a total of 15 stages and a one stage per day cadence. But there’s 24 hours in a day and it’s for sure possible to find enough time to hike two stages in a day… Right? Unfortunately, I got injured on the third day, and I had to cut my trip short and head to the beach earlier than planned - ahah.

There is no right or wrong way to hike. If you prefer to spend the afternoon at the shelter to enjoy the view and connect with fellow trekkers, then go for it. If you want to complete a trek as fast as possible for the challenge, then that’s okay too. As long as you take necessary precautions and avoid taking inconsiderate risks, you should do it your way and enjoy the experience.

Planning ahead

Early options

When planning a trek, the first decision to make is which trek to go on. My advice is to choose a trek based on what appeals to you, whether it’s a region you want to explore or a particular type of trail, such as coastal or high-mountain terrain. You can find plenty of inspiration online or in dedicated books.

Equally important is deciding how you intend to trek. There are two primary considerations:

• Will you carry a tent, or will you sleep in shelters?
• Will you bring your own food and stove, or will you eat in shelters and resupply daily?

These decisions will affect the size and weight of your backpack, as well as your budget. For my treks, I have mostly slept and eaten in shelters, with the exception of a 3-day autonomous epxerience in Belgium.

Getting informations

Once you’ve made the initial decisions about your route and how you’ll be trekking, it’s time to dive into the data. This is where you can really start to get a sense of what you’re getting into and what to expect along the way.

Here you want to get an exhaustive knowledge about the stages of the hike, including their length and difficulty, where you can rest, sleep, and refuel. Those kind of information are typically found online or within dedicated books such as “topoguide”.

When I planned my GR54 trek, I made sure to compile detailed information for each segment:

• Time estimate
• Distance
• Elevation gain
• Elevation drop
• Availability of shop and shelters
• What official stage is the segment part of

One key resource that I highly recommend is finding a GPS trace of the trail. This can be a lifesaver when you’re out there on the trail, making sure you stay on course and don’t get lost. I typically look for GPX files, which can be loaded onto a smartphone app like Gaia GPS for easy reference.

It is also the moment to consider alternative paths on the trek. Often, there are unofficial trails that lead to breathtaking and less crowded views. For instance, passing by the Incudine on the GR20 or Eychauda’s Lake on the GR54 provide great experiences.

Another important aspect to consider is the gear and clothing you will need, depending on the region and the season during which you will hike. This will ensure that you are well-prepared for any weather conditions you might face.

Once you have all the necessary information, you’re ready to pack your bags and hit the trail. Deciding how flexible you will be during your trek is also crucial. You can follow a set schedule, decide which stage you will complete each day, or keep some freedom. If you plan to be autonomous, you can live day by day, eating and sleeping whenever and wherever you want. However, if you plan to rely on shelters, you may want to book in advance to secure your spot.

Personally, I enjoy the flexibility of a serendipitous trek. I start my day early and keep moving until I feel like stopping. Knowing how long it takes to reach the next possible stop using the segment plan is super helpful. But remember, some shelters might be at full capacity, so be sure to understand how likely it is to happen by getting some feedback online or directly calling them.

With all this knowledge in hand, you’re ready for your adventure. Get ready to embrace the unexpected, discover new paths, and make unforgettable memories.

My tips for actually hiking

Merino all the way. When it comes to hiking clothes, merino fabric has been a game changer. That magic material comes with several benefits, the most important for me is that it does not stink over time and you can thus wear the same clothes multiple day without issue. I use a single T-shirt on my treks. Merino is also a great thermo regulator, keeping you warm when needed and allowing your body to breathe, avoiding wetting your clothes. Merino wise, I typically bring with me socks, legging, short sleeve T shirt, long sleeve T shirt, sleeves and briefs.

Volume over speed. To hike longer distances, it’s easier to hike more than faster. I typically trek 10 to 13 hours a day and avoid hiking in the dark. I start my day just before sunrise, usually around 4-5 AM, and stop when I feel like it or at sunset. I rarely run and I am a relatively tall and heavy hiker, thus slow uphill. I do typically run downhill though. This way of trekking usually makes me 2 to 3 times faster than standard trek planning.

Store water on the front. To lighten my backpack, I use a hydration vest backpack. These types of backpacks have pockets on the straps to store water bottles, distributing the weight and allowing easy access to water. Check out Ultimate Direction for some great options.

It’s about the people. While the views are incredible and the sense of achievement is fulfilling, the people you meet on the trail are what make the experience truly unforgettable. Even though I usually hike alone, the encounters I have had in the mountains have been unique and meaningful. Whether it’s chatting with the shelter guardians or fellow hikers, the conversations are always great and caring. I recommend sleeping near official shelters and sharing your dinner with others. If you sleep in a shelter dormitory, make sure to bring earplugs!

Then, here are some more practical tips:

• Arrange your different clothes in compression packs for efficient backpack organization
• Use toe socks to prevent rubbing between your toes and avoid blisters
• Set up elastic laces for effort-less shoe slip on and off. They also provide a strong yet flexible hold
• Hook some carabiners to your backpack, it will be handy in different ways, such as letting your clothes dry while hiking
• Opt for foldable hiking poles for quick and easy dismounting
• Have a running belt to store a few stuff around your waist, with straps on the back to hold poles
• If you are into music, bring wired earbuds that don’t require recharging, use guiche clips to lock the cable onto your backpack straps
• Carry mixed nuts with you for snacks. Get sweet and salted mix
• Red tiger’s balm will help your muscle recovery

If you’re interested in specific brands and models, check out what was in the backpack of my latest trip.

Are you sufficiently trained?

The safest way to know is to try. Load up your backpack and lace up your hiking shoes, then hit the trail near you. Start with shorter hikes to gauge your endurance and determine whether you enjoy the experience. If you want to estimate how much distance you can cover daily, try measuring how much elevation you can handle within an hour. If you live in a flat area, head to the gym and try out the stair climber. Then combine your estimated distance and elevation paces to determine how fast you can hike a stage.

Hiking requires more than just physical endurance; it also demands a functional body and resilience. If you’re not used to hiking regularly, you may develop injury if you go too hard, such as a join inflamation. However, some discomfort is OK, e.g. a blister is not a trek-stopper, if you do not mind too much. So it’s important to differentiate between discomfort and pain, to ensure that you’re not pushing yourself too far and putting yourself at risk of injury.

For reference, here’s a look at my own 6-months volume of walking and running over the years.

I began running somewhat regularly in 2013 and ran my first ever 10km in 1 hours and 7 minutes back then. Unfortunately, I could not find my tracking data for that period. The above chart is based on Strava data, which I started using in 2019.

Summer 2018 was my first attempt at the GR20. During that one, developped a tendinitis and gave up. Two years later, for my second and successful attempt, I had reached 450 km volume over 6 months. Keep in mind that these figures aren’t comprehensive, but they should give you an idea of my level of training.

Remember, hiking can be a challenging and rewarding experience. But it’s important to ensure that you’re physically and mentally prepared for the rigors of the trail. So take your time, start small, and build up your endurance gradually.

See you in the moutains

Thank you for taking the time to read my tips! Of course, recall that what works for me may not work for everyone, and there are many different ways to approach a trek. Take what you find useful and adapt it to your own needs and preferences. If you have any further questions, please don’t hesitate to reach out to me directly.

I learned Dynamic Web Development by playing Soccer

Sports-wise, I am a late bloomer. Indeed, I started practicing regularly when I was about 16 years old. I use to just not enjoy it, and I cannot blame my parents as they tried to enroll me in many occupations such as soccer, rock-climbing, table tennis and fencing. Anyhow, I now appreciate athletics a lot, I began my journey by getting in shape with weightlifting, then endurance training, i.e., jogging. Retrospectively, I commenced to love physical activity when I acquired a minimal fitness level, which was somewhat awkward to land. As a kid, I did not push my comfort zone too much.

At university, I kept my fitness activity to fundamental exercise, which eventually led me to trail running, as you may know. Upon saying goodbye to my student life, I investigated team sports. Soccer specifically, as I had been a fan for several years.

Starting late

Thus, I looked for soccer games and joined some via via. I realized that ball touch is difficult, and requires practice. I would often contest with folks who had been playing for many years, typically back in high school, and bringing something to the team was… hard. Luckily, I had my cardio and strength to leverage. I also rehearsed some drills alone.

After a while, I wanted to join a club and I heard about a Brussels-based amateur league, called ABSSA, meaning Belgian Association of Saturday Sports. Eventually, I found a team to play with, and my immediate next step was to block all the games’ times in my calendar, so I would make sure to attend and not double book anything else.

Scraping ABSSA’s website

The league’s schedule is available on its website, so I could have just browsed it to log all my games but why do it manually when you can automate it? Playing a bit with the inspect tool of my navigator, I noticed that the internet site is scrapable, i.e., its information is extractable. I also noted that to have all the data about a specific match, I would need to visit several pages, one for the game time, one for the field name, and another one for the field location. Something annoying to do manually but simple to code.

Hence, I opened my favorite editor and made a prototype to:

1. Scrap the info on the website
2. Combine them into an iCal file that I could import in my agenda

I quickly got a suitable result with Python and a couple of libraries. This tool started as a Jupyter Notebook then moved to structured classes and was just refactored in a data pipeline using Kedro. This project is open sourced on GitHub.

Publishing those calendars

The advantage of automating a process is that it is scalable. I had developed that calendar maker for my team, yet I could now also generate the schedules for every single team, as the logic was the same. And there are 258 of them. How should I make those calendars available to everyone? I could share a drive folder with all the iCal files but the user experience would not be great and it would limit its reach to minimally tech-savvy players.

What if I build a simple website where a user would search for its team, see all its fixtures and have the option to save them in its agenda? Hmm… Should I develop it, it would look something like this…

Project based learning

As I shared in my very first blog post, I decided to get into web development, to add such new string to my bow. This personal blog is an initial step into this direction, but it is fully static, meaning that its pages’ content will always be the same, regardless of the action you do on it. For my calendar sharing app, I would like the search bar to have an autocomplete function. Its suggestions will depend on what the user types in it. Thus, a static website is not the way, and it is the occasion for me to learn a new tech!

I am about to create a web app with client side rendering using JavaScript. Different frameworks exist for such need, and your first call is to select the one you will apply. I like to make my decisions data driven and, luckily for me, there are large surveys being rolled out on JavaScript technologies by the State of JavaScript. Let’s look at the latest results for front-end frameworks:

For my first front-end development experience, I would like to master a technology which is standard and well appreciated. Filtering out Solid and Svelte which are too recent, React is ranking top.

Hence, I went on and learned React. I started by its tutorial then dived straight into trying to create my project. A week later, I am happy to share my very first dynamic web app: calabssa.be.

Looks pretty much like the original design, right? Check the code on GitHub.

Welcome Twitch!

While embarking on my React apprenticeship, I reckoned that it would be cool to document my learning process. How would I tackle acquiring a new tech skills? To that end I decided to live-stream my desktop while working on the project and think out loud. I have been streaming on Twitch and it’s fun, feel free to follow my channel as I intend to keep broadcasting! Family and friends were my main audience but some strangers passed by too, leading to some unexpected exchanges.

I started by sharing my web development effort but it was pretty hard to explain what I was doing while deeply focusing on understanding the new technology. Once I landed a first working prototype for my web app, I went back to the calendar creation to improve it. I fully streamed refactoring it to a Kedro pipeline, and the cool thing with lives is that you get a replay, which I combined and published on YouTube:

Lessons learned

Developing websites is exciting as you get a direct feedback. Indeed, as soon as you hit save on your editor, you will see your web page update itself and view its result live. This is very different from data projects, my main skill. For those, the outcomes of your work are typically less tangible as they are more research oriented. You will end up with a number, which could have a big impact on your business eventually, but it’s not a live web app where each line leads to a specific element on it.

For this project, I employed React with Tailwind to handle the page layout, the former managing the JavaScript and the latter the CSS. The development experience was nice because a single file contains all the components of a web page, i.e., HTML, CSS and JavaScript. Hence, it is convenient as you do not need to modify multiple files to rework a page. This is different from the approach I used for this personal blog, where I rely on Express and EJS and have to deal with them in separate files.

A pitfall of such rapid feedback loop and self-contained development is to organically update small bits of code indefinitely. Indeed, it is very tempting to add just one line or fix this tiny element, but if you keep working this way, you will soon end up with spaghetti code. Even for my small app I spent some time refactoring it into several modules, or “components,” for a more sustainable software. The most technical of you may check this out in the code difference of this commit.

Should you want to join the web development movement, it is important to be mindful of the various mechanisms happening simultaneously. Indeed, your web page will be the result of different renderings (client, server) and technologies (HTML, CSS, JavaScript). Thus, to stay on top of the dynamics in your website, make sure to know what piece of code is responsible for what. For example, I was surprised that most of the animation in my website where handled by its CSS, while the JavaScript would just update some hard-coded like value.

When learning, I also encourage you not to leave stones unturned. If there is something you do not understand, it is easy to just keep moving and avoid it, but I would argue that taking the time to grasp it will pay off in the medium run. It will allow you to build a strong knowledge foundation, which will eventually enable exponential learning growth.

All of this because of soccer! It’s a great example of project-based learning, i.e., a teaching method in which students learn by actively engaging in the real world and personally meaningful projects. I believe that such approach is the most effective, find something you want to do, and do it. It will be more efficient than memorizing theoretical concepts or working on nonrelatable homework.

Cheers!

Ultralight trekking the GR54: 185km in 5 days

I have been trekking for several years now, below is a short diary of my latest hike on the GR54, during summer 2022.

Day 1

Though night, difficult to wake up, slightly nauseated. I get up at 4:30 am and leave at 5:15 am without eating, I have 3 hours planned to the refuge 500m higher. I start with the headlamp. I end up having my first feeling of middle mountains, but I am exhausted. Sometimes I close my eyes in spite of myself.

Finally, I reach the magnificent shelter of l’Alpe in 2 hours. There, I live again. I eat a breakfast that makes me feel good and I discuss with Clément who will sleep at more than 3000m tonight, before taking the GR54 again. I pass the col d’arsine totally reboosted, the mountains are breathtaking, there is a white river.

And 3h later, here I am at Monêtier-les-Bains, close to the Serre Chevalier ski resort. Either I follow the GR54 with ski lifts in the middle, or I take an alternative path to check out a lake at 2000+. As I am faster than my estimations, I go for the latter and do not regret it!

Villar-d’Arêne - Vallouise: 45km 2050D+ 2515D-

Day 2

The night was short once again, for some reason I cannot find good sleep. 4:30a I wake up and eat some stuff I bought at the grocery store, as I was staying in a town.

5:15a, I leave the hotel and start hiking. The day begins with 8km close to a road. It’s a steep walk though, I see car passing close by. I am a bit tired, should I just hitch hike couple of kms? No, I push it until about 7a, eat a cereal bar and wander into “Le parc national des Écrins”. It’s beautiful. Still pretty cold but beautiful.

The hike until Pré de la Chaumette is long and uphill, offering different sceneries. On the way, I cross an unmanned shelter with a couple of autonomous hikers. They join me on the journey, with their big backpacks. Minutes later, they win terrain on me, eventually taking over. How is it possible, with so much weight on their back, am I too slow? Can I even make it?

At the summit, we share some sausage and then it’s downhill for 900m. Galloping, I arrive at the shelter first, eat “oreilles d’âne” and a blueberry cake. It’s 1pm, originally, I planned to stop here, but is still have time. Next shelter, estimation is 6 hours and 45 minutes. I’m typically a bit faster… Let’s go!

I close the stage, I am the very last one to start. It’s two mountain passes in a row before reaching the Vallonpierre shelter, next to its lake. With passes, each visit opens to a mystery new view, stunning every single time. After the 2nd pass, I am not there yet, actually it’s three passes in a row!

6pm, I arrive at the shelter. It’s full, I have no booking… Will I spend the night on a bench? Thankfully, the guardian sets up a tent for me. There’s lamb for dinner, and a beautiful sunset on the lake. See you tomorrow!

Vallouise - Refuge de Vallonpierre: 37km 2826D+ 1697D-

Day 3

4:30a, finally a good night of sleep! Is it related to the friendly dinner yesterday? The fresh air breathed from the tent? The awesome location at 2200m? Most likely all of the above.

5a, I head out to the shelter for breakfast, the guardian prepared it for me, with warm coffee in a thermos. Today, there are three other early birds. It’s a family on a hiking holiday, they prepared their trip carefully checking out IGN maps.

Half an hour later, I refill my water bottles and go. It’s a steep downhill to the valley. The early lights make for a vivid landscape. There’s a shelter midway. I notice a guy with a T-shirt colored to the Belgian flag. He and his friends are from Tournai. He tells me that the water here is not drinkable, good that I refilled.

7a, I’m at 1000m, on my way to Villar-Loubière. It’s flat. For 12km. Next to a road. Yesterday during dinner, I meet a group of 4 friends, with trail running-ish style. Turns out they are wrapping the GR54 counterclockwise, in 5 days, bivouacking. Meaning carrying a tent and food, for about 12-15kg. Pretty impressive. Thus, they came the way I’m heading to. They decided to take a bus for this flat section. Should I also skip it hitchhiking? What about the alpine alternative going through l’Olan, that’s not flat? And actually, will there be room for me this evening?

As I understand that yolo-ing every decision on this trek is harder due to accommodation high demand, I decide to book a bed in a gite 3 steps later, with a total estimation of 13 hours. And what about the next day? And the day after? Well, at the end of day 3, I will be at end of step 9, on a total of 14. Wait, am I about to complete this trek in 5 days? 175km 11kD+? I don’t know, but let’s book as if. Alea jacta es.

Eventually, I decide to 1. not take a car, part of my drive for those hikes is to reconnect with nature, and hopping into a car would compromise that. And 2. to just walk the flat, I’m about trek the most I ever did in 5 days, while not being at my best physical form, compared to previous years. Let’s just recover a bit on that portion.

11a, I start the ascension toward le refuge des souffles. It’s 900m uphill, but quite easy. Indeed the terrain is non-hazardous and not too steep. 2 hours later, I’m enjoying a cheese and ham omelette before heading to le Col de la Vaurze. It’s superb! The rock is dark, like a volcano, or what I think it looks like.

3p, it’s time for downhill. 1300m downhill. The first 150 meters are extremely steep. I would have preferred to hike that uphill. Anyway, after some time, the dark scree become dirt with rocks. I could run down here, but I do not. I prefer to preserve my knees, I’ll reach the shelter early enough anyway.

7p, after settling in and doing a bit of clothes hand washing, the kind of thing normal hikers do every day when you don’t arrive dinner-30 minutes, we start eating. It’s a long table, with 16 plates. The guardian is a passionate cook and spent all the afternoon simmering sausages and rice. It’s delicious. Those shelter dinners are always a nice time, everyone is so caring. The typical opener is “so you’re on the GR? When did you start?”. At some point, I answer those questions, so… I’m 60% in and started 3 days ago. This morning I was at Vallonpierre. As I am talking, I notice more and more guests becoming silent and looking at me with big eyes. I share some tips and tricks on how to pack lightly, avoid injuries etc. Of course I know my way of hiking is a bit out there, but it is not crème de la crème either right? What I do in 5 days, Kilian, Lambert or Jim would do in 30 hours. To them I was still somehow extraordinary, everything is relative.

Refuge de Vallonpierre - Désert-en-Valjouffrey: 36km 1660D+ 2659D-

Day 4

5a, it’s a solo breakfast today, the guardian left a small goodbye note, that’s nice. Today is going to be a cool one. The plan is to reach Vénosc, two mountains away. 1000 meters up, 1000 meters down and repeat.

In between, there is Valsenestre. Actually going there is a small detour, which I would like to avoid because it does not have much interest. Problem: how do I get enough water for those two mountains at once? The estimated time is 10h45 and my water capacity is 1.7 liters. I anticipated that, a fellow hiker gave me purification pills, allowing to drink river water. This discussion led to a heated debate about the need to purify that glacier water. As I want to minimize risk, I’m going to use the pill for sure.

5:40a, I’m heading toward the Col de Cote Belle and as I look in the opposite way, i.e. where I came from yesterday, I see a lone bright point at the top. I’m not the only one starting my hikes with my headlamp. The walk is nice and I feel great. No more sleepiness, I guess that hiking while fasting is not for me.

9a, I reach the other side, I am way faster than estimates. I barely drink, as it is not too warm yet. Still, now is the time to refill. I see a river, but it smells so bad I decide not to drink it, even purified. I see on the map that I should cross the same river but more upstream. Good. As I start to hike the second mountain, I see a Shepherd’s house and, guess what, there’s a water source next to it.

11a, I’m close to the top, the Col de la Muzelle. It’s super steep, in endless laces, on a dark-colored material. I see several people at the top, chilling with a view. Soon it will be me. I discover the other side of the mountain, there is a huge lake, the lac de la Muzelle. Beautiful. After a mandatory selfy and some nuts, I head down. The lake becomes bigger and bigger. There are several falls due to melting ice, flowing into the lake.

12:30p I just joined the shelter, there were some people attempting a swim in the lake, up to knees mostly. I share my lunch with two Dutch. They think I’m a cyclist as I wear Sutro sunglasses. Ahah, no, I just enjoy the full field of view. Now it will just be downhill to Vénosc, I order a last coffee to enjoy the view up here.

3:30p that’s it for the day, too bad I pre-booked, I might have kept walking until 6. The next possible stop is 12km further so it would have been a push. I will be sharing my room with 4 other hikers today.

7:30p we have dinner and, as usual, we chat. There are two girls hiking together, they met in La Réunion. They ask me if I do trail running race. I don’t. But they do. La Diagonal des Fous actually. Woaw. It’s one of the most popular races, with the UTMB. She completed it in 59h, can you imagine? “Simon, with the way you trek, don’t you want to do trail running races?” I do, actually, stay tuned!

Désert-en-Valjouffrey - Bourg d’Arud: 25km 2170D+ 2521D-

Day 5

4:30a, I sneak out of the dormitory, trying not to wake up anyone. I usually pack my bag before sleeping so that I can leave the room smoothly, as soon as my alarm fires. For breakfast, I make coffee using a percolator, feels good after several instant coffee days. Today, I will follow the C part of the GR54C. It’s an alternative path of the trek; basically it is 20km shorter for the same total elevation.

5:30a, instead of heading toward Bourg-d’Oisans, I start hiking to the Lac du Chambon, right next to Mizoën. To do so, I go around the Pied Moutet, a 2400 meters tall mountains, in the Les Deux Alpes skiing domain. On the way, there are several equipped sections, with small ladders and chains to facilitate some perilous paths. Nothing like the GR20 though, I almost never had to use my hands on this trek. The only exception was when hiking the Col de la Grangette on the first day, but which is actually not part of the GR54. Several kilometers later, I pass by an unmanned shelter. There is a small table and two beds. It would be really cool to spend a night there, maybe one day, when I’ll do autonomous trekking.

10a, I’m arriving on top the Lac du Chambon, it’s huge! The water is sky blue and I can see some water sports happening on it. After a small section on the road, I cross the lake on its dam, it looks super high. I one shot a cola and a ice tea at Mizoën and head toward the Plateau d’Emparis, my final rise. On the way, I see that I can go check out a small lake, Le Lovitel. But I see no lake, it is completely dry. Apparently, the whole massif is dry this year, it is usually way greener.

2 hours later, I make it to Les Clots shelter, although I am hungry, I decide not to have lunch here but higher up instead, to enjoy the height just one last time. On the way up, there is a waterfall, actually multiple coming from the same rocks. The way all the water flows is beautiful. On the southern side, I see two glaciers, their white coats are impressive.

2p, at the top of the Plateau, I realize it is the last time I go above 2000 meters. Thankfully, I’m on a “plateau”, meaning I will have some distance to travel up here before getting down. I take my break at the Mouterre shelter. I eat a crepe with an awesome view on the glaciers. Even though I am not hungry anymore, I decide to have a dessert, just to enjoy the moment a bit longer. The cook offers me a local digestif, genepi, made from herbs growing there.

3:30p, eventually, it’s time to go. I start hiking toward le Chazelet, and then, La Grave, my final stop. It’s still 800 meters downhill. My feet are pretty tired, so I take my time. I cross the path of some hikers going upward, it reminds me of myself, just 4 days ago, when I was starting the trek. The way down goes into laces, there are several shortcuts, but signs warn us not to take them, as they erode the landscape.

6p, I am at La Grave. It’s over. 185km, 5 days. I did not expect I would have been so fast. I think that the terrain was easier than on my previous hike, enabling a faster pace. The GR54 is beautiful, and steep, but not to technic. At the gite, I share my dinner with a former physics researcher, now retired. We discuss rock climbing, hiking in the Pyrenees, quantum physics and cross-country skiing the GR20. I still love those sharing moments.

Tomorrow, I will be leaving for Brussels, by bus and train. As usual, this adventure fulfilled me to the fullest. The views, the effort, the judgment calls, the people, the fresh air… I cannot wait for the next one!

Bourg d’Arud - La Grave: 41km 2488D+ 1962D-

Wrapping up

Should anyone have been inspired by my experience, you can find what was in my backpack here and my itinerary just below

The moving time is estimated by Strava.

So… This was my third large trek, completed in 5 to 6 days. I plan to share the knowledge I acquired doing them, from preparation to day-to-day hiking. You can expect to find a new post about this some day.

In the meantime, you can also check-out those highlighted stories for more views on the GR54 trek.

Cheers, Simon

How I became McKinsey Belgium’s first Data Scientist

At the end of my master in Computer Science, I knew pretty well what I wanted to do. Leverage my technical skills to solve business problems. To that end, I enrolled in a couple of management-related electives during my final CS years. After this glimpse, I decided to deepen my business knowledge, and I pursued an extra master, the Master in Management from Universidad Carlos III de Madrid.

New country, new school, new learning. Today, I am glad I went for it. On a professional side, I acquired knowledge about strategy, finance, marketing …, but also many soft skills. Indeed, the teaching approach was centered around group work and presentations. This was a great training for me, it allowed to practice communication and team management skills, with students from different backgrounds and countries.

On a personal side, I discovered a new city, met new friends and I could cheer for my favorite soccer team, in their unique former stadium, the Vicente Calderon.

Kicking off my career

With a strong background in Computer Science and Management, I was ready to leverage my technical skills to solve business problems. Back then, my flatmate was working as a Data Scientist for a Big 4. He explained to me how he was helping TelCo companies to optimize their shop presence using their internal data and Machine Learning algorithms. What he was doing seemed awesome to me, and I went on reading about the Data Scientist roles, entitled as the sexiest role of the 21st century by the Harvard Business Review.

Now I just had to find a Data Scientist job! Easier said than done… Indeed, for such position, a company will grant a lot of responsibility to the employee, as he or she will have a strong impact on its business operation. It can lead to great efficiency improvement, or to a waste of time, depending on how competent is the Data Scientist. Therefore, firms will look for medior profiles, especially if their analytics team is small. Hence, I struggled to find a junior friendly position, which would give me the required experience, the chicken and egg problem.

Resiliently, I eventually discovered a great opportunity at a local boutique consulting company, specialized in Google Cloud. There, I worked on many interesting projects. From adjusting the marketing campaign of a leading car manufacturer thanks to machine learning generated customer segments, to increasing the understanding of a NPO’s fan base by developing an intelligent social media listener tracking an evolving list of keywords and hashtags built on natural language processing.

I also meet great colleagues, such as Charles, who became a cloud expert and recently launch his own consulting venture, 10/10 would recommend!

The McKinsey datathon

As I was cruising within my Google Cloud Data Scientist role, for about half a year, I received a LinkedIn message from a partner of McKinsey Belgium. He explained how they were creating an analytic hub in Belgium and that they were organizing a datathon, i.e. a Data Science oriented hackathon, to which I was invited.

Of course I accepted. At the time I was casually active on Kaggle, thus aware of the kind of contest I could expect. I was looking forward to being part of such an in-person, Kaggle like, competition. Moreover, during my Spanish stint, I had heard of the great place that McKinsey is, and how it could give me access to even better and unique opportunities.

Couple weeks later, I joined the event with another 100 participants and … I won first prize!

I was so excited about the win that, on my way home, I punctured a tire of my car, because I was on the phone and not focused enough on the road - ahah.

On top of the cash prize, I was also invited to the interview process at McKinsey. It was a unique opportunity, as, in theory, they would not hire junior profiles to kickstart their Belgian hub. But since I demonstrated my abilities, they made an exception for me.

I pursued the recruitment procedure, and ended up getting an offer. Back then I wrote about this whole experience and it was shared in a blog post.

The interview

A couple of traces of this interview still remain online, a tweet and an Instagram post. On the other hand, the direct link to it is not active anymore.

The text below was originally published in October 2018.

Why did you decide to participate in McKinsey Analytic’s March hackathon?

It started when I received a message in my LinkedIn inbox. The mail was a solicitation to participate at a McKinsey Hackathon. It also said that this contest was taking place in the process of building the Advanced Analytics hub in the Benelux. The invitation was very appealing to me for two reasons:

• I was very curious to discover how McKinsey is handling the analytics competences;
• I love to work on Data Science problems, for me it is similar to solving a brainteaser or an enigma. Thus, I decided to apply for the event and, a few days later, received a confirmation.

What was the most fun and interesting part of the hackathon? (actual Data Science Challenge)

I enjoyed the context of the hackathon. We got a dataset containing records of patients and the goal was to predict their survivability. For each record, we had physical information about the patient such as age, sex, weight; as well as the disease and the treatment. I liked to work on this topic because it is very impactful. Indeed, the forecast could help doctors to give a better medication to the patient and eventually save lives.

What did the recruiting process look like after the hackathon?

After the event, I planned a call with a recruiter from McKinsey Advanced Analytics. During the exchange, I could ask my questions on being a Data Scientist at McKinsey and I received the information about the hiring process.

Then, I have been in contact with a consultant from the local office, in Brussels. We talked over the phone about the Advanced Analytics department and did a small simulation of a case interview. The consultant gave me feedback and advised me on how to prepare myself for the real interviews. The last step before the actual interview was a coaching session at McKinsey’s office. Other applicants and I met in person with a consultant. Together, we solved a full case and discussed about the personal experience part of the interview.

Finally, I had my first round of interviews a few weeks later. Each round consists of three interviews, with three different interviewers. Two of them were general business interviews and the last one was focused on analytics. Later the same day, I received positive feedback and the invitation to the second and last round of interviews. After two weeks, I went back to McKinsey’s office for the final round of interviews. The process is the same but you meet more senior people from the company.

Overall I enjoyed the recruitment process, the discussions I had with employees from McKinsey were interesting and it felt more like an exchange than a one-way interview.

Who do you recommend participate in this kind of hackathon?

I believe that anyone with a genuine interest in analytics is a good candidate for the hackathon. Indeed, you will have the opportunity to meet with fellow passionate people, test and strengthen your Data Science skills and finally, discover the careers that McKinsey can offer.

What did you enjoy the most about the McK Hackathon? (whole event)

During the event, I liked to discuss with employees from McKinsey. After the competition, we could connect with other participants and consultants from McKinsey. I enjoyed having an informal talk with them and being able to ask my questions on the practice. I learned about the different types of roles within the Advanced Analytics practice and how a typical project is handled.

How did you prepare for it?

My preparation was a day-to-day effort. Being a hired Data Scientist, I expected the hackathon to require similar competences as the one I practice in my daily activities. In order to put all the chances on my side, I focused my learning in three areas:

• I made sure to keep up to date with the latest algorithms in Data Science;
• I put an emphasis on being proficient with data wrangling and preparation;
• I went over a few Kaggle competitions to have an idea about the different types of challenges that exist in Data Science and to look for best practice.

What would you suggest to future McK hackathon participants?

For the hackathon I attended, efficiency was the key of my win. Indeed, the length of the challenge was limited to four hours. Thus, we could not afford to lose any time. My recommendation would then be to master your basics. One should be quick at identifying the type of algorithm one is going to use. For this, make sure to take into account the goal of the task (classification, clustering, …) and the type of data (image, tabular, …). Be also prepared to deal with some data engineering and cleaning. For example, if you decide to work with Python, as I do, be ready to wrangle data with Pandas.

The goal is for you to have the time to try different models and work on feature engineering.

Final thoughts

Eventually, I did join McKinsey as a Data Scientist, I was followed by two colleagues the next month. I went on and worked with many great people, over 10+ countries and the rest is history.

As a side note, I was not the first ever employee to work with data science at McKinsey Belgium, but I was the first Data Science Fellow, i.e. a client facing consultant, who leverage his knowledge in that field to implement AI models, from inception to production.

Swapping my company car for a bike, 1000 km later

With a new job, comes a new package, and, an important aspect of the said package is mobility.

In Belgium, for some legacy reasons, offering a company car to its employees is fiscally interesting for the employers. Meaning that, for a given cost, a firm can lease a car for its staff which would be more valuable to the worker than its equivalent in cash. Of course, this depends on individual preferences, but many companies will standardize their package and include a car as part of their typical benefits, presupposing it will matter to the most. Although this car should have a professional motivation, the employee can enjoy it for private use too.

When joining a startup as the first employee, such assumptions were not made, and I had the chance to tailor my package.

So, do I need a car or not?

As with many decisions I make, I usually start by defining and structuring the need, likely a habit from my past as a consultant. In this case it was pretty easy, I just had to look back at what my private automobile use was. It boils down to two main purposes:

1. Traveling for holidays
2. Grocery shopping

Living in a dense city, I am used to moving with soft mobility means for small journeys. Indeed, I personally dislike traveling by car if that means dealing with traffic and frequent start stop. I also often notice that the automobile is slower than free-floating e-bikes or e-scooters when including parking time. Finally, it is always easier to come back with a cab after a couple of drinks when your automobile is waiting for you at home!

It was clear that I could find alternatives for my two remaining car needs. And thankfully, my whole family lives in the same city as I do, so visiting them is feasible with soft mobility too. Many of my friends who moved to Brussels for studies and work will travel back to their relatives, and it’s usually way faster by automobile than public transport.

Thus, what’s the plan?

• Traveling for work: bike or public transport
• Traveling for leisure: bike or free floating scooter
• Traveling for groceries: bike with some cargo space
• Traveling for holiday: rent a car or go by train
• Anything unforeseen: rent a car

A bike, but which bike?

Opting for a bike as your main mobility mean is nice, but you still have a couple of details to nail down, after some thought I went for:

• An e-bike, or bike with electric assistance, for speedy movement without sweat
• With front suspension, to cope with Brussels paved roads
• Fenders, so that mud won’t splash on my back
• A luggage rack, to carry extra bags e.g., for groceries
• Manual transmission, as I feel smooth automatic transmissions are not there yet
• A removable battery, to allow charging it anywhere
• A frame adjusted to my size

With such requirements, trendy e-bikes such as Cowboy and VanMoof were disqualified as they do not fit half of them. Bye-bye the startuper cliché, credit to @loule.blou.

I went ahead and investigated the Canyon brand, which I had been eyeing for some time, especially the Endurace model to get into weekend road cycling. For this everyday bike, the Pathlite was ideal, except for its price which was above budget. Then I discovered an outlet was available, selling discounted bikes due to minor imperfections. Problem, new bikes were added to the outlet sporadically, without notice…

I started by checking out this outlet occasionally, but soon realized that I should automate this monitoring. And so, I did. I quickly assembled a script which would request all bicycles on the outlet, scrap them, save them in a CSV table and send me a daily email with new bikes matching my search criteria.

I developed this whole script in Python and deployed on Google Cloud Platform using Cloud Functions, Storage, Pub Sub and Scheduler, the code is open source.

After a couple of weeks, my dream Canyon bicycle was still too expensive, even discounted up to 10%. You can check out all the bikes on sale I scrapped in this CSV table if you want to make some stats. Hence, I changed my tune by looking at other bike brands. I learned about Cube, apparently having a strong quality to price ratio, partially due to a low marketing spending, like OnePlus back in the days. Their Touring Hybrid One 400 checked all the boxes, including the cost!

1000 km later

I recently passed the 1000 km mark on my new bike and as I have been cycling almost daily for several months, I thought I’d share some of that wisdom with you.

Gear yourself for bad weather

I biked the whole winter and enjoyed it, you just need to get the equipment. It’s the same for resort skiing, you would not appreciate it without feeling warm and dry. So what do you need? First are rain jacket and pants. Any kind will do, in Europe you’ll find cheap options at Decathlon.

Then I recommend merino gloves and leggings. The gloves will prevent your hands from freezing due to the wind, the leggings will keep your legs warm outside and fresh indoor, thanks to the magic properties of merino. I get my merino from IceBreaker because that is what Mike Horn uses and what Reddit suggested.

You also want to grab a pair of waterproof gloves, on top of your merinos when it’s raining. And last but not least, a windshield, because, getting 25 kph droplets in the eyes makes driving difficult. Your windshield can either be a pair of large frame glasses or a cap.

Get a bike theft insurance

If you are considering commuting by bicycle, then you are probably living in a dense city, which comes with many advantages but also some inconvenience, such as more frequent larcenies. If your bike will be your main means of transportation, you must be able to park it and keep your mind at peace. With the rise of two-wheel commuters, insurance company started to offer bike theft protection. For about €120 a year, your bicycle will be fully reimbursed if stolen. Although losing your ride is still a pain, at least you’ll get a new one without repaying it.

However, make sure to read the conditions which apply for the coverage. I was surprised to learn that my bike would be insured if locked up to any hoop in the street, but not in my private garden when locking the wheel to the frame only. I had to install a specific fixed point in my yard to follow the requirements. I could have attached my bicycle on the road just in front of my place, but I did not want to fill in a theft report every couple days, as grinding a padlock is a matter of minutes nowadays.

Biking makes micro traveling nicer

Since getting my e-bike, I have been making many more commutes with it. I used to rent free-floating e-bikes for some journeys, but the use cases were limited. Free floating options are pleasant yet will often be slightly unadapted, e.g., with some minor however inconvenient mechanical issues. As I selected just the right bicycle for me, this restriction went away, and boosted my usage.

Today, I confirm it, traveling 25 min by bike is nice, 25 min by car is okayish, 25 min by public transit is annoying. I think I am particularly averse to public transports because I get motion sickness in buses, if I could read a good book while in the metro, I guess it would be another story.

If you feel unsafe, you are

Road sharing is still a work in progress, at least in Brussels. Some drivers are careful, others are not. Therefore, if you think you’re at risk of accident, then you are. To cope with this, make sure to apply the traffic regulations, yes that means yielding priority to cars even if you are going full speed downhill or struggling uphill. On the other hand, do take your priorities and do not be afraid to claim some space on the road. You are not supposed to stick to parked cars, especially if the street is in poor conditions there.

Thankfully, biking infrastructure is improving in major European cities, but never underestimate lack of knowledge, distraction or bad will from automobile drivers. Always be ready to protect yourself, e.g., by braking if a car suddenly turns and cut your path. Although you have the priority, it is better slowing than crashing. Hope for the best, prepare for the worst.

Thanks for reading and see you soon, e.g., at the next Critical Mass 🚴.

I have been coding for more than 15 years, but never in JavaScript

As a long-time computer scientist, I know that the field has many topics to offer. I started my journey back when I was starting my secondary education, i.e., when I was about 12 years old, by learning how to code in C++, following an online tutorial on OpenClassooms, formerly known as “le SiteDuZero”.

Back then, I was just trying to create some games after a friend told me he had developed an in-terminal app where the user has to guess a magic number. It worked as follows:

1. The computer generates a random number, the magic number
2. The user types a number as a guess
3. The computer tells if the magic number is above or below the current guess
4. Repeat 2 and 3 until the user finds the magic number

Much fun you would say, and actually it was. I learned about the basic of programming, i.e., variables, branching and loop statements, etc. Trying to solve some toy problems by coding was an intellectual challenge I enjoyed, and still do! I guess it is similar to answering riddles or filling crosswords.

Anyway, I went on and joined a CS program at my local university, eventually graduating from a Master with a focus on AI and optimization.

But…

While I did learn many great concepts, I always felt that I was lacking a specific CS-related skill, and what a skill. It allows you to directly interact with end users and present all your cool results. You guessed it, I am talking about web development.

Before college, the further I went with web development, was to explore another tutorial from “le SiteDuZero” about HTML and PHP, but then I moved to automating some games I used to play, using AutoIt. This story will be for another time.

During college, I was fairly proficient with Python and thus, I thought that developing websites with Django, a Python web framework, would be pretty easy. I even had an idea, I was going to make some kind of streamer schedule, where each streamer can define when he will be streaming and subscribers can then have a combined agenda of all their favorite streamers’ schedules. Unfortunately for me, I was quite lost in front of all the web concepts, such as ORM, templating, routing. I was trying to skip too many steps at once.

The switch

A couple of months ago, at the time of writing this article, I changed jobs. I joined a Belgium-based startup, Wequity, with the ambition to use finance to promote sustainable investment. This mission would materialize in a ESG (Environment, Social & Governance) dashboard, where end-users can monitor how well companies are doing in those dimensions. In order to make that happen, we built a pipeline to fetch articles and social media posts from the web and extract the ones about ESG. Great! We had those data, but we were struggling to allow our end users to access it with a good experience. So we hired a lead full-stack developer.

Once again, he would be able to create an awesome web app with dynamic graphs. I was amazed.

Talking with my colleague, he told me how quick it is possible to become proficient with web development using JavaScript especially with Node.JS. Tools I had never used before, although I know they are standard in the web industry.

JavaScript sometimes has a bad reputation, I never really understood why. Arguments often seems pretty subjective such as “it is an ugly language” or “it is error-prone”. I decided it was time for me to make my own opinion! And to that end, I built this website with in JavaScript running on a Node.JS server.

Getting started

• Hey, I’m going to create a JavaScript website!
• Oh, cool, which framework will you be using?
• …?

Of course, there are several ways to build “a JavaScript website”. When developing a web app, there are challenges which you will encounter and that have been faced by many before you. That’s how frameworks emerged. Basically, it’s a collection of pre-built codes which you can use for typical web tasks. There are a lot of frameworks, have a look at an overview here.

My colleague advised me to start with simpler frameworks and recommended me a Udemy course, which is apparently a gold standard. So I have been following it for a couple of months, learning about the Express framework, which had been used by 81% of web developers in 2021, according to the overview I shared above.

You can follow my progress on this GitHub repo, and about a quarter through the lessons, I now have better understanding of web development overall, not only about JavaScript and Express. Concepts which did not make sense in Django now do, and I feel as if I could go back to it. This is thanks to the great material from the instructor, which explains practically how to use it but also what happens under the hood.

However, I will stick to JavaScript. Indeed, it is widely used, and I believe it is not by chance. I heard that you can become extremely productive with JavaScript for web development, and that is one of my goals.

This project

Okay, I am learning about JavaScript, what do I do now? Hey, remember that cool blog from Duarte? Let’s make one too! Any resemblance is purely coincidental.

I met Duarte when I joined Wequity, he is a great Machine Learning Engineer, and was helping the project. He is also a blogger, so naturally I asked him how he made his website. He explained to me that it was actually a statically generated website, using the so-called Jamstack.

I decided to reinvent the wheel and built something similar, but using what I am learning. Of course, this website is unlikely to be efficient, but that is not the point. The point is to practice, get better, and share my journey along the way. Please feel free to check out my coding skills, as I made this website open source.

Which brings me to the end of this post. So what can you expect? Well, basically anything I feel like sharing. Currently, my areas of interest are mainly tech and sports. I already have a couple of ideas for future posts about those, in the meantime, you can already check out some videos I made. I also intend to make this website evolve, maybe some day I will transform it to a single page app, or try the latest and greatest frameworks.

Thanks for reading this introduction and see you soon!