I came to code from the operations side.

Sport, then an injury.

I was a competitive athlete from childhood through college. I went to the United States chasing a scholarship, got injured, and came home.

What survived was the habit. Structure, repetition, and a process you can point at when the result is bad. I still work that way. It is why I would rather hand you a number a computer can check than an opinion you have to trust.

Insurance, then national infrastructure.

After the injury I read Financial Management at the University of Buckingham. That is what put me into finance, and finance is what put me into operations, at NN Insurance, 1,500 people.

Then the digital parking and ticketing platform for the City of Budapest. It launched on time and served millions of users, with no launch failure.

Cross-functional teams of up to fourteen people across time zones. Finance, healthcare, government and software. The PMP came out of those years, and it is still the only certificate I carry. Everything I have built since is checkable without one.

The pattern repeated everywhere. A company grows faster than its systems. Tools get bought and half-configured. The team works hard and the infrastructure underneath is held together by willpower. I became the person who fixed that.

I stopped specifying systems and started building them.

I studied software and DevOps engineering mid-career, on a long hands-on programme rather than a second degree. Then I started shipping.

Hundreds of n8n workflows in production. More than ten self-hosted n8n installs built from scratch on Docker Compose, PostgreSQL and Caddy. Next.js, TypeScript and React on the front. Python and Node behind them. MySQL and Postgres underneath. Swift when the answer is an iOS app.

In the same year I rebuilt a video production pipeline in code and shipped a sales operating system for a fifteen-person business. The method was identical. Map the estate. Find where the money leaks. Build the smallest thing to stop it. Then decide whether more is worth paying for.

I run a department and I still ship.

I lead the AI and Automation department for a US e-commerce brand doing around USD 1 million a month. Three engineers report to me. I set the architecture and the delivery standards, so the pieces fit.

I wrote the technical assessment we hire against. It measures how well someone directs AI inside an existing architecture, not which language they know. Twelve candidates are running it now.

I own unit cost, account ownership, credential rotation, offboarding and security posture. Nobody owned API spend before I arrived. A finished ad on the pipeline I rebuilt now costs USD 2.34 per minute.

Two positions I will not move off.

Decisions live in code, judgment lives in prompts, and never the reverse. If a rule matters, it is a number a computer checks, not a sentence a model is asked to respect. Where spend is the risk, I put a ceiling and an alert on it before switching anything on.

And recommendations from someone who profits from the work are worth less. So when a finding points toward hiring me, I say so, and I name the alternative if you go elsewhere. I told one founder to put his budget into ads instead of the system he asked me to build. I told another his vendor was about to write logic in the wrong place entirely. Both are still clients.

Client data stays on client infrastructure. You get the export, a credentials map, runbooks and a walkthrough. I am not a dependency you have to keep paying.

Outside of work.

I live in Hungary with my wife, our twin daughters Hanna and Bella, and a giant schnauzer called Hector.

I still think like an athlete. Structure, discipline, consistency. The competition is with myself now: build a better business today than yesterday, and be more present with my family while doing it.