Hands-on data lead. I build trustworthy AI analytics: semantic layers, guardrails, evals.
- I build: data stacks ready for agentic analytics: dbt, semantic layers, Snowflake, DuckDB/MotherDuck, Python and Dagster, plus evals to check the agents.
- Also worked with: Airflow, AWS (S3, DMS for CDC, Lambda), Iceberg, GitHub Actions, Docker, Claude Code.
- I've built: the data stack from scratch, twice (Door2door, nebenan.de), and led the data team at nebenan.de.
- Impact at nebenan.de:
- led the personalised notifications end to end: click rates doubled, daily active users up about 24%
- cut Snowflake ingestion costs by more than 80%
- moved the personalisation pipelines to MotherDuck: 30% lower compute costs, near real-time
- Now: honest-agent, open-source evals that check whether your AI analytics agent got the right answer from the right sources.
- Writing: The Data Lead, on data infrastructure, analytics engineering and leading data teams.
Work with me: I've gone freelance, building data foundations and AI analytics hands-on with companies. Get in touch through The Groundlayer or on LinkedIn.