Building production ML systems means owning the full picture - from experiment design through deployment, monitoring, and reliability. That's where I operate.
At HelloFresh I've shipped production ML end-to-end across two major domains: first as co-owner of recipe recommendation and cold-start ranking across multiple international markets, then leading addon demand forecasting while supporting box forecast, meal-swap forecast, and audits optimization on the side.
I'm most effective when I can move between data engineering, modelling, and deployment in the same week. I work daily with Python, PySpark, Polars, Pandas, LightGBM, MLflow, and Flask; and own the full lifecycle from experimentation through model training to production jobs on Databricks via Asset Bundles.
Lately I've moved into a new layer: building LLM-powered developer tooling with LangGraph and the Claude API that makes the whole team faster.
510+
Pull Requests
20+
Markets Served
27
Repositories
4+
Years @ HelloFresh
Rhythm is my native language - I play bass and drums, and follow live music as much as I can.
From sourdough to aged cheese - I care about where ingredients come from and what they become.
Studying the language, reading the stories, watching the shows - Japan has been a long fascination.
Using AI as a creative tool - generating art, designing stickers, and exploring where code meets aesthetics.
See my work on PixivWhether it's a new city or a long road - moving through the world is how I reset.
Rarely have the time, but when I do - only the ones with a story worth telling.
