AI engineer focused on agentic LLM systems — multi-agent orchestration, RAG, and the evaluation that makes them reliable. MSc in Computational Science & Engineering at ETH Zürich, currently a Research Assistant at the IDEAL Lab.
I'm most interested in the reliability side of agentic AI: making multi-agent systems behave consistently in the real world, not just in a demo.
- EngiAI — Hierarchical multi-agent LLM framework (LangGraph/LangChain) for engineering design: RAG, MCP, HPC orchestration, and an evaluation benchmark across multiple LLM backends. Accepted to ASME IDETC/CIE 2026.
- prusa-mcp — A Model Context Protocol (MCP) server that lets LLM agents control Prusa 3D printers.
Python · LangGraph / LangChain · RAG · MCP · PyTorch / JAX · C++ · Docker · HPC / SLURM
Daily driver: Claude Code. Certified in Anthropic's Introduction to MCP and Advanced Topics.
Extending EngiAI at the IDEAL Lab — exploring how agents stay reliable under non-deterministic model behavior.

