Work together

Piotr Obolewicz

I build production AI systems — currently in healthcare, where a confident wrong answer costs more than no answer. Alongside that I take on a small number of engagements a year, on the problems my own system already forces me to solve properly.

What I work on

Retrieval & evaluation. RAG systems measured against a real eval set — grounding, citations, and defined behaviour for when the answer isn't in the corpus.

Production AI systems. The part that decides whether a pilot survives: timeouts, idempotency, cost per request, degraded mode, observability — and the human decision boundary written down rather than assumed.

Technical AI governance. EU AI Act readiness from the engineering side: which systems you're the provider of and which you merely deploy, and what to fix first. Your lawyers own the legal position — I work next to them, not instead of them.

Looking for

Companies putting AI into production — moving from a promising prototype to a system someone is accountable for.

Healthcare partners to pilot MedBot — triage and evidence-grounded recommendations, tested against real workflows.

Early-stage teams that need an engineer who has shipped both regulated backends and AI systems — as a collaborator, not a vendor.

Full-time roles, selectively — senior backend (Java), tech lead, or architect. For the right problem I'd rather be on the team than next to it.

Background

The services behind Poland's national road tolling system. Bank operations integrated across twenty-one microservices at Santander. Credit-risk platform work at ING Hubs. Audited, regulated, expensive to get wrong — which is where the instincts come from. Full history on LinkedIn.

Tell me what you're building — a paragraph is enough.