About
AI Engineer and AI Systems Consultant. Based in Berlin, Germany.
Currently an AI Engineer at Prospecto, building national-scale procurement intelligence pipelines across 20,000+ government sites. Before that, five years of data and automation consulting for teams in Germany and the Nordics.
Most AI systems fail quietly: a retrieval step returns the wrong chunk, an agent hands a bad state to the next step, a fine-tune degrades behavior nobody is testing for. I build the layer that catches that before it ships: schema gates between agent handoffs, drift detection on legal text, eval sets pulled straight from production failures instead of invented test cases.
Three of these are public. Compliance Agent screens policies against explainable YAML rule packs, backed by LangChain + ChromaDB retrieval. LegalDrift catches semantic drift across legal-document revisions using four statistical tests over embeddings, not just a word-diff. DiffDDL diffs database schemas as ASTs instead of strings, so an agent can propose a migration without silently dropping a column.
I write about the mechanics under the hood: why RAG decays silently, why hallucination is architectural, not a bug, why a five-agent pipeline of 90%-accurate agents nets 59% end-to-end accuracy.
Experience
Distributed pipelines aggregating procurement data from 20,000+ government sites; NLP/LLM tender classification and entity resolution across suppliers, buyers, and tenders.
Python ETL and automation (n8n, Zapier, Power BI) for recruitment, sales, and HR workflows at Siteimprove, Hedvig, ABC Labs, and BIIOC.
Validation pipelines integrating ESCO/ISCO accreditation frameworks with Danish and Swedish national registries.
This site: written in Obsidian, built with Next.js, deployed on Vercel.
I'm open to consulting engagements and full-time roles.