Unboxable in Tech: 30 Years of Being the Wrong Shape

Source: Dev.to

This is a submission for the 2026 WeCoded Challenge: Echoes of Experience

Every time a recruiter or staffing agency reaches out, the conversation ends exactly the same way.
“Your profile is really interesting… but honestly, we’re not sure which box to put you in.”

At first I took it for polite awkwardness, then for a back‑handed compliment. Today I take it as the most honest diagnosis anyone has ever given me.

Thirty years of coding, modeling, debugging at 3 a.m. because the business problem hadn’t been understood upstream. Thirty years of building systems that work — and the industry keeps looking at me like an IKEA cabinet without instructions.

What if the problem wasn’t me?

Act 1 — The TO7 and the Accidental Revelation

First contact: a Thomson TO7 in a middle‑school classroom, mid‑eighties. A slightly mad math teacher showed us how to make characters dance in BASIC on a cassette tape. For a thirteen‑year‑old, it was Narnia with GOTOs.

Three years later I was on a pure humanities track—Latin, Greek, German, philosophy, endless essays. No keyboard in sight. Yet when I finally enrolled in a computer‑science school, the virus was still there, lying dormant.

The school was a different world entirely: COBOL, CICS, JCL; mainframes, large organisations, formal methods. MERISE for data modelling, Warnier for program structure. Rigor, hierarchy, documentation.

First internship: digitising the library of a middle school on a machine built around an Intel 8088, two 5¼″ floppy drives, and GW‑BASIC. A modest setup—nothing like the systems we’d been taught to build.

Very quickly it became obvious:

The program is never the real problem.
The real problem is badly structured data. Loans, returns, authors, copies—if the relationships aren’t defined correctly from the start, nothing holds. So I modelled first, before writing a single line of code.

The method doesn’t depend on the machine.

At twenty, without quite knowing it, I had chosen my side: systems thinker before coder.

Act 2 — The Patchwork Years

What followed was a joyful, improbable patchwork.

• Architecture firm – quantity surveying in dBASE III+, an HP printer with no manual. Two weeks deciphering the jumpers on the back like a Chinese puzzle, because you had to understand the machine before you could ask anything of it.
• Regional postal authority (when Chronopost was still in pilot mode) – small applications to track shipments and complaints.
• Industrial cleaning company – OS/2 and Lotus Symphony, automating invoices and schedules.

Each time the same ritual: dive into the actual business, model the flows, structure the data, then code. Understand first. Build after.

No permanent position. No defined job title. No linear career. A way of working that held up—but had no name in the industry’s classification grids.

Act 3 — Paris, the Staffing Agencies, and the Cultural Clash

In 2006 I tried the “normal” circuit: Paris, open‑plan offices, staffing agencies, Pierre & Vacances, Photoways, on‑site contracts. I wanted to see what it looked like from the inside. I saw.

The full Agile rollout: Redmine tickets, daily stand‑ups, retrospectives, velocity points. Three to four hours of ceremony per week for, in the end, ten minutes of actual thinking about the system.

I delivered. But internally I was screaming:

“You spend more time filling in burndown charts than figuring out why the client is losing 15 % of their leads.”

Entire sprints were spent slicing user stories to the millimetre—without ever asking: Will this data model hold up in five years? We velocity, we velocity, and we’ll patch it later.

It wasn’t Agile itself that was the problem. It was what it had become: a collective reassurance ritual that consumed exactly the energy and focus I reserved for the actual work.

One exception

Logic‑immo (2008‑2009) – hired through a staffing agency. I walked into the interview for a Zend Framework role and walked out twenty minutes later, hired for something else entirely: an Oracle‑to‑MySQL migration, PHP in CLI mode, one non‑negotiable requirement—data integrity first, execution speed second. No trendy framework. No buzzword bingo. Just a roadmap.

Someone had looked at what I actually did, not what was written on paper. Half on‑site, half remote, data at the centre, airtight scripts. The kind of work where I could give my best.

It was rare. Rare enough to matter.

I held on until 2011, then packed up and left. In 2012 my daughter was born. It wasn’t a retreat—it was a conclusion.

Act 4 — Building on My Own Terms

Leaving didn’t mean stopping.

• Regional classifieds site – Zend Framework, Smarty, dynamic generation and management of over a hundred business sites with custom routing. Ten months, solo.
• B2B matching platform – time‑slot constraints and human/automated arbitration; algorithmic optimisation, smooth interface, performance that made the previous system embarrassing.
• Other projects – chosen, where no one asked me to use a particular technology—just to get a concrete result.

Then dev.to, writing in English, and projects nobody had commissioned.

• RFC 2324 – the HTTP protocol for coffee pots, an engineers’ joke from 1998. I implemented it seriously in Python with a raw asyncio server, because it was the right excuse to talk about what protocols actually say about a technical culture.
• My own satirical RFC – HTBMCP/1.0, “beer‑mug control protocol”, port 1414, numbered 1516 as a nod to the Reinheitsgebot.
• AJC Bridge – a WordPress plugin that syncs content to static‑site generators via GitHub. It didn’t exist in the form I needed, and it was the right problem to solve.
• CV as a graph – an analysis of my own résumé as a graph of relationships rather than a timeline, because a linear CV can’t represent a trajectory like mine.

None of these projects have a job description. None of them are “full‑time” in the traditional sense, but each is a piece of the puzzle that finally lets me fit myself into a box I designed myself.

ack React Node AWS. And yet they all start from the same place: the real problem, not the résumé to fill.

One last example. I submitted AJC Bridge to the official WordPress.org plugin repository. After weeks of waiting, the reviewer assigned a generic slug that overrode the plugin’s name — no explanation, no interest in my objections, citing a hypothetical trademark issue on a word that belongs to no one. I withdrew the submission. The plugin is now distributed directly via GitHub Releases.

Same mechanism. Different arena.

Act 5 — The Same Job, Thirty Years Later

I no longer code full‑time. I work as a classroom aide.

I’m not going to pretend it’s the job of my dreams. Being a classroom aide means being underemployed by choice — or rather, by refusal. Refusal to keep fighting a system that will never know what to do with me. Refusal to sacrifice my region, my daughter, my way of working, to fit into a mold that was never cut for me.

It’s a compromise. Lucid, deliberate, and sometimes exhausting.

But it’s mine.

What I do with these kids still looks strangely like what I used to do with systems: observe without assumptions, understand how things actually work rather than how the documentation claims they do, adapt, cobble together solutions that work for this particular person—not for the box someone is trying to put them in.

My methods are just as disruptive here.

The WeCoded challenge celebrates marginalized voices in tech. I’m not a gender minority. But I’ve worked alongside enough colleagues who are to recognize the mechanism: it’s not that you don’t fit in the box — it’s that the box was built to exclude you. Whether the stated reason is gender, background, credentials, or simply a way of thinking that makes people uncomfortable — the outcome is identical.

Marginalization isn’t always visible. Sometimes it’s just — unclassifiable.

If you’ve ever been handed that line — “great profile, but not really in our boxes” — listen carefully: you’re not broken. Their classification grid is just too narrow.

The industry loves ultra‑specialized components. It has little tolerance for people who see the whole system and dare to say: “this is going to collapse in two years if we keep going like this.”

You’re not unclassifiable. You’re just illegible to a matching algorithm.

And no, you don’t need to wait for them to learn to read.

When everything collapses under the weight of technical debt and broken promises — the unclassifiables are the ones they come looking for. Not to thank them. Just to fix it. And then they forget about them, until the next crisis.

So build your space now. Not while waiting for permission.

They won’t ask for your box. They’ll just ask: “Can you fix this?”

Yes. I can.

And so can you, probably.