Bartosz Mikulski | AI & MLOps Engineer

I tried to draw a bicycle using a mouse, with photos of bicycles open on a second screen.

For a long time, I wanted to live in Berlin.

This article is a text version of my talk, "The Failures That Don’t Crash: MLOps for AI Agents," presented at Berlin Buzzwords 2026.

An MCP server is a user interface for an AI agent.

There’s a graveyard of AI projects that never made it past the pilot phase.

“Why can’t we just put all of our documents into the context window?”

You’ve been there.

How much time have you wasted in the frustrating black box of debugging a bizarre AI hallucination?

What if you could sniff out AI hallucinations without building a custom test set or babysitting outputs?

I used to think embedding speed was just background noise.

How would a mismatch between the user query language and the source document language affect RAG?