About me

Hi! I'm Philipp — a Senior Applied Scientist at Zillow. I build machine-learning products end to end: from the data and the models through to the services and APIs that ship them.

I work primarily on home valuation, property data, entity matching, and recommendations, directly with product managers and design. My focus is on what several products need in common: data and models that serve more than one of them, and infrastructure that is already there when the ask arrives.

Much of what I've learned about ML lives in the gap between a model that works in a notebook and one that works for months without you even noticing.

I'm drawn to tools and services somehow connected to the physical world — cities, spatial data — and to visualization and small tools other people can actually use. I got here the long way around — first an architect, then an urban planner, then a data journalist, and now an applied scientist. I've long stopped treating those as separate careers — it's one way of working that's just had several names.

Before moving to NYC, I taught Urban Informatics at university and Data Science at school, and co-created Branchpoint, a non-profit project teaching parametric design — perhaps one day I'll get back to it.

I keep a few pet projects and write here now and then. Some places to start:

• Agate — an Obsidian plugin that draws charts from your notes.
• pyCombo — packaging a research algorithm so strangers can install it, three times over.
• A short foray in data journalism — what we built, what it taught me, and how it ended.

The full list has the papers, talks, and older work.