About me

Hi, I'm Krystian. I live in Wrocław, Poland, where I'm a husband, a father and an AI practitioner, and I try to keep them in that order. The AI part keeps trying to move up the list.

These days I lead teams that take generative AI out of the demo and into the real world, where people ask awkward questions like "why did it say that?" and "was it saying the same thing last month?". I got here by a winding route: EU research projects, then leading a machine learning team at Nokia, and now enterprise GenAI at EY. There I was technical lead for a RAG assistant used by more than 100,000 people. That is a lot of people who will notice when it gets something wrong.

The demo is the easy part. What interests me is everything after it: how you know the system is good before you ship it, who gets the call when it isn't, what the model was actually looking at when it answered, and how to stop your engineers from spending the whole day grading the AI's homework.

Lately I've been drawn into knowledge foundations: semantic layers, knowledge graphs and ontologies. I'm also curious how teams work once AI writes much of the code and humans are left reviewing it.

I write here for people who have already seen the demo and now have to live with the system. You'll find opinions, the methodology behind them, the occasional mistake I made so you don't have to, and very few predictions. Predicting AI is a great way to look silly six months later.

If you're wondering how generative AI could genuinely help your business, beyond a chatbot and a press release, I'm happy to talk.