About

I help companies turn AI bets into shipped products — from the first roadmap conversation through the code that goes to production. Most AI consultants pick a lane: strategy or engineering. I do both, which is what makes the difference between a polished plan that stalls and a feature your users actually trust. Fifteen years of shipping software, the last few of it spent watching where AI features quietly fail once real traffic shows up.

By day I lead engineering at a large enterprise software company. By night I run two ventures under the same banner: huntech.dev , a community platform for Hungarian developers, and hodi-group.hu , my holding for personal products. Right now I'm experimenting with Claude subagents and cowork patterns in production — testing what autonomous-coding workflows look like when there's real traffic and revenue attached.

Here's what I've seen repeatedly: agent demos work beautifully. Agent features in a real codebase rarely do. The gap isn't the models or prompts. It's full-stack engineering — schema design, latency budgets, error paths, the infrastructure that turns a clever idea into something people actually use. That's the work I'm built for, and it's what I take on through consulting.