As Tech Lead on ŠkoPilot at Škoda Auto, I architect and ship an enterprise multi-agent assistant serving 30,000+ internal users — owning the design and writing the code, not just directing it.
Five years of hands-on engineering across AI, backend and architecture — with the last 1.5 leading production AI at enterprise scale. Measured, not claimed.
Multi-agent systems with tool/function-calling, MCP integrations, and context management across multiple model providers — built on a forked, extended Google ADK.
Long-lived eval frameworks that track LLM output quality over months, not just at launch — plus production RAG with document-parsing and contextual memory.
Solid, maintainable foundations — from event-driven backends to high-concurrency JVM platforms serving over a million sessions. Built to scale and to last.
Long-form system and platform designs for production AI. Each one states its assumptions up front, argues the trade-offs, and commits to a decision — because the reasoning is the deliverable, not just the conclusion.
One multi-use-case agent platform, the seven-phase lifecycle that governs it, and the argument for which use case to build first. Buy the runtime, build the governance.
Grounded Czech intent-to-catalog mapping as strict structured output, at 1.2M conversations a day inside a 5–10 s p99 budget — where the schema, not the prompt, makes hallucination impossible.
Agentic systems, evals, AI-native developer tooling. Open to lead & architect-level conversations — always up for a good one.