Writing

Notes on building machine-learning systems people can use.

Long-form thinking on recommendation systems, experimentation, research design, and production data science.

Written to make technical choices, trade-offs, and failure modes easier to reason about.

Reading shelf

Ideas I return to.

Selected work on recommender systems, experimentation, research software, and production machine learning.

Prefer the work itself?

Explore projects where the methods meet real decisions.

View selected work