Be concrete
Start with real examples and observable behaviour before reaching for a broad metric.
About
I’m Stanislav, a computer science student interested in making AI systems easier to understand, test, and trust.
I’m in the final year of my BSc at the Faculty of Informatics, Masaryk University. Alongside my studies, I work part-time at Lexical Computing, where I’m building an automated LLM testing suite as my bachelor’s thesis.
I’m drawn to the practical side of AI: finding where language-model systems fail, designing evaluations that reveal something useful, and turning the results into better engineering decisions.
I like taking an unclear question, shaping it into a small experiment, and iterating until the result is understandable enough for someone else to use and improve.
Working principles
Learn in public
Write down the decisions, the surprises, and the useful bits that are worth revisiting.
Own the details
Reliable systems come from respecting the unglamorous edges as much as the central idea.
Where I’m heading
I want to keep working on AI systems where evaluation is part of product engineering, not an afterthought. I’m especially drawn to small teams, meaningful ownership, and technical problems that reward careful thinking.