Hey, I'm back.

Mon, March 30, 2026 -
Hey, I'm back.

2026, maybe my last year of dev as I knew it 10 years ago.

2016 — My beginnings in dev

2016, I land in this world through web dev, in a Parisian startup that was the “shazam for celebrities” (Reminiz, later became Context).

Basically, I had to work on an internal app to annotate celebrity photos, store them in a database, which was then used to recognize celebrities on live TV feeds.

It was cool, I learned the basics of web dev on the job, straight up full stack mode. No time to mess around.

Originally, I have a background in robotics and machine learning (anyway, that’s not the point here).

Shoutout to Reminiz’s CTO from that bygone era @Antoine Biard, who trained me in this craft in the best way possible:

  • 1 hour of vim every day (yeah I know, the guy was hardcore)
  • touch typing sessions during the day, to reach 120 wpm and ship code as fast as possible.

I’m telling you, it was a crazy time. I was learning something new every single day. I loved it.

Doing everything, learning everything

Then the years went by, I spent 4 years at that company, where you had to handle everything yourself, between newbies and passionate devs.

The whole stack, from design to code, to infra, monitoring, basically all of that on the back of a 4-person team. You had to learn everything.

When I say everything, I mean everything:

  • creating / managing a k8s cluster (not with today’s tools, we’re talking kops, no managed EKS, no GKE, none of that).
  • developing the apps (internal and client-facing), architecting them properly
  • deploying monitoring tools and managing errors, alerts, etc
  • meeting client SLAs (well we didn’t have that many clients, but enough for it to be a challenge).

And we made it. We sweated it out, but we got there. And damn it felt good to learn, to feel alive through your work.

OK… I’m exaggerating a bit but it was truly enriching.

What came next?

After that glorious era, I switched jobs, looking for new markets, wanting to see new things.

Going freelance, doing your craft for a direct client -> completely different. Less dopamine, but more cash.

That career kept going, and it still does today (thank God). But for over a year now, one thing has changed.

The elephant in the room

The elephant in the room

Let’s talk about the elephant in the room: genAI. Not Kling or Nanobanana (nobody cares about that here), but coding agents. BOOM.

We won’t go through the whole history, but basically we went through several stages:

  • Github Copilot -> garbage, janky autocomplete, sometimes funny (go try typing “Hey baby, wanna see my ” in copilot and wait for the autocomplete. (It’s probably been fixed, don’t try it)).
  • Cursor and other early AI-assisted IDEs -> first decent results, features were shipping week after week, but it still wasn’t magic.

And then…

Claude Code

Sweet Jesus (relax, I’m French, it’s a cultural thing), what magic! This tool was a unanimous hit among my colleagues and friends after a weekend of testing.

The thing ships code (oh it generates it alright, the 120 wpm are useless now).

If you give it a solid foundation, you (almost, calm down) don’t need to type code anymore. You express a request, and the agents execute. Fascinating. Terrifying.

Where does that leave us?

And there, as a software engineer, you start asking yourself a lot of questions about your role, your career, and the future of this profession.

So yes, we have some semblance of increased productivity, because it’s physically impossible to type code as fast as these agents generate tokens.

But, if you take your craft seriously, you review everything. Absolutely everything. Even if you have tests.

Now, I’m not saying I do it every time (cognitive overload) but you have to, to avoid surprises and especially to be able to understand and explain what was implemented and how.

So this isn’t an article about “for or against coding agents”, it’s not 2025 anymore. The answer to that question is crystal clear -> you can’t be against it. The industry won’t let you.

And so… the backlash, and this is the key point of this comeback: desire, passion.

Many developers I know are in this spot: not knowing if they love this job anymore, or at least if they still love it in its current form.

If we go back to my 2016 experience, the hours of grinding, learning, reading docs, binging hours of YouTube tutorials (shoutout to the Indian devs for the content) to acquire skills and be able to use them the next day at work…

That era is sadly over, I believe.

If you’re reading this and you’re just starting out in this field, don’t hesitate to use AI to learn faster, but don’t fall into the trap of Plan -> Approve without at least trying to understand what’s happening.

The comeback

After that terrible old man lecture, let’s get to the title of this post: my comeback.

Well.. my blog only had 3 posts before this one, but they date back to 2021… I had to come back.

Why now?

Because I feel like I have the time to transfer my knowledge, my projects and my experiences. 2026, Claude Code and I’ll tell you what I told it to build.

Next article: a project that’s already done and working, which will serve as the foundation for the next ones: a small kubernetes cluster with raspberry pis.


See you soon,

Adem.