Articles
Writing
Notes on analysis, ML systems, and making data legible — the thinking behind the work, not just the results.
May 12, 2026·7 min
Shipping a ranking model to production
The notebook was the easy 20%. Here's the other 80% — serving, evaluation gates, and rolling back without drama.
ML in ProductionMLOps
March 28, 2026·5 min
How to read an experiment that looks flat
A headline-flat A/B test is rarely a no-op. The story is usually hiding in the segments — if you look without fooling yourself.
A/B testingStorytelling
January 15, 2026·6 min
Making data legible to people who don't love data
Most analyses die in a slide deck. A few change a decision. The difference is almost never the model — it's the translation.
StorytellingVisualization