Piotr Obolewicz

Fukuoka, Japan · Warsaw, Poland

I build production AI systems. Right now that means healthcare triage and evidence-grounded decision support — a domain where a confident wrong answer costs more than no answer at all.

I come at it from the engineering side rather than the model side, and that has shaped the whole project. The hard problems turned out to be retrieval quality, eval design, failure taxonomy, permissions, and defining exactly where the human decision boundary should sit.

Designed for European constraints from the start: EU AI Act, GDPR, human oversight, traceability, and explicit decision boundaries.

Current questions: what evidence should be required before clinical deployment, how much autonomy should an AI system have, and where should human judgment remain mandatory?

Senior Java engineer by background. Building first, incorporating second.

If you're putting AI into production, or you're in healthcare and what I'm building maps onto a problem you have — I'm open to working together.