the dev work.
Most companies don't need more AI. They need someone who can make it actually work.
What He Does
He builds AI systems for real operations, and teaches the people who have to live with them. Not demos. Not decks. Working tools that hold up on a busy day, and a team that understands them well enough to keep going after he leaves.
The unusual part of the résumé is the useful part: he spent twenty years running kitchens before he ever shipped software. High volume, thin margins, real people, no room for a system that only works when everything goes right. That's the same room most AI projects die in — and he's been on the receiving end of enough bad tools to know exactly why.
The Work, Two Ways
Implementation. Find the handful of places AI genuinely earns its keep in your operation — usually the documentation, the training, the repetitive decisions nobody enjoys — then build it, ship it small, and prove it works before anyone's asked to trust it.
Education. Teaching is the half most rollouts skip, and it's the half that decides whether anything survives the first month. Training cooks was the job for two decades; training a team on new tools is the same job with a different piece of equipment. Plain language, hands on the thing, no jargon for its own sake.
available for consulting, contract, and full-time roles
Vibe Coding, With The Boring Parts
Everything on this page was vibe coded — described in plain English and built in conversation with AI, rather than hand-typed line by line. That phrase usually means a weekend demo that falls over in front of a real user. These are live sites with real traffic, a playable game, and internal tooling that runs unattended.
The difference isn't the prompting. It's everything around it. A pipeline that refuses to report success until it has re-read the live site and confirmed the change is actually there. Checks that run before anything ships. Small releases, so a bad one is a bad hour instead of a bad week. Anyone can vibe code something that works once — the skill is the discipline that makes it hold.
That skill transfers directly. It's the same thing most teams are missing right now: not access to the tools, but a way of working that turns fast output into things you can actually run a business on.
How He Works
AI as a collaborator, not a vending machine — the kind worth arguing with, because one that agrees with everything is one you can't use. Ship the version that helps today, then grow it. And nothing is called finished because the code got written; it's finished when someone has watched it work. That last rule sounds obvious and is the one most projects quietly skip.
On the Line
Everything below was built this way. Two of them are sites you're welcome to judge — you're standing in one.