Petru Arakiss

I build AI systems that survive contact with the real world. While others chase the latest models, I focus on the boring stuff that matters: reliability, observability, and making probabilistic software behave predictably at scale.

I lead AI engineering at Atlax 360. 19 years of building software taught me that the hard part isn't the prototype. It's making it reliable under load, bad data, and tight budgets. I spend most of my time on the unsexy stuff: error handling, monitoring, cost controls, and guardrails that prevent 3 AM incidents. Also run the frontend platform because good AI is pointless if nobody can use it.