I'm Tomáš. I measure before I claim.

Student based in Brno, Czech Republic. For the past months I have been building AI systems and small tools with an AI pair programmer, running my own tiny web studio on the side.

Download CV (PDF)Email me

What I bring

Hands-on work with model APIs beyond chat: function calling, JSON schemas, probability-based classification, model fallback.
Tests before claims: scenario scripts, word error rate benchmarks, blind human versus model tests, linters on model output.
A written log of decisions with reasons and dates, including the directions I dropped.
Failures reported with the number attached.

What I can't do yet

Most of my code is written by Claude. The ideas, decisions, tests and fixes are mine. Writing it fully on my own is the skill I'm working on.
TypeScript: basics only. It is next on my list.
None of my agents has run in production with real users at scale.

Skills

What I work with

LLM APIs

Groq, NVIDIA, OpenRouter, TypeSafe through OpenAI-compatible SDKs. Function calling, JSON schemas, prompt injection defence.

Speech

Whisper (local and hosted), Cartesia TTS, Czech text normalisation, word error rate measurement.

Everything around it

Python, Playwright, Apify, scraping, JSON, browser DevTools, Obsidian as a decision log. Czech native, English.