002 / On Stage

Speaking

Talks on AI engineering workflows and document tooling. Slides, recordings, and photos where they exist. Organizers: grab bio and photos from the media kit.

OCT 2026

4Developers Katowice · Katowice · Vienna House Easy by Wyndham

Grading the Ungradeable: LLM Evals for Real Products

Most developers treat evals as the unit tests of the LLM era. They are not: unit tests check code, evals check behavior, and when "correct" is a rewrite or an edited document rather than a label, pass/fail stops working. A field report from running production evals at SuperDoc: deterministic checks, golden datasets, LLM-as-judge, evals in CI, and the traps along the way.

October 8, 2026Event Watch recording

Recording

JUL 2026

RLabs Builders Meetup · Warsaw · Google for Startups

Why Your AI Engineering Setup Is Probably Wrong

AI can already generate front-end code — the hard part is turning that speed into reliable production workflows without losing control of quality, context, and design consistency.

July 21, 2026Event Watch recording

Part 1 · 34 min

JUL 2026

WAWTech + Summer · Warsaw

Why Your AI Engineering Setup Is Probably Wrong

Most teams bolt an LLM onto an existing workflow and call it AI engineering. A walk through the failure modes — no evals, prompt sprawl, unversioned context — and what a setup looks like when it survives production.

July 18, 2026Event

JUN 2026

VUS Kraków: Tech Mixer · Kraków

Your users lie. Their prompts don't

What people say they want from a product rarely matches what they type into it. How we read real prompts as product signal and let them reshape the roadmap.

June 23, 2026Event

MAY 2026

Python Pizza · PJATK, Warsaw

SuperDoc SDK: a modern alternative to python-docx

python-docx has carried Python's DOCX story for a decade. A look at what a modern document SDK does differently — fidelity, collaboration, and AI-ready editing.

May 9, 2026Event Watch recording

Recording