What I'm reading right now, what's in the queue, and what I recently finished. Updated roughly monthly. Books I rate highly enough to mention here have stuck with me past the third chapter.
Currently reading
-
Regression Modeling Strategies — Frank Harrell. Slowly, chapter by chapter. The treatment of splines is changing how I think about feature engineering on continuous variables. This is the book for anyone who wants to be careful with regression in a clinical or high-stakes setting.
-
The Anthropic interpretability papers — Anthropic's circuits team. Currently working through the recent sparse-autoencoder work; planning to write up notes once I finish.
-
The Pragmatic Engineer — Gergely Orosz's newsletter. Best signal-to-noise for "what's actually happening inside good engineering teams" I've found.
In the queue
-
Causal Inference: The Mixtape — Scott Cunningham. Going back to the diff-in-diff and synthetic-control sections with worked examples. Free online.
-
The Book of Why — Judah Pearl & Dana Mackenzie. Third re-read. New things stand out every pass.
-
Designing Data-Intensive Applications — Martin Kleppmann. Re-reading the streaming chapters as I think about evaluation pipelines.
-
Statistical Rethinking — Richard McElreath. The Bayesian textbook I should have read in grad school.
Recently finished
-
The Mom Test — Rob Fitzpatrick. Short, opinionated, immediately useful. Should be required reading before any user research.
-
Working in Public — Nadia Eghbal. Why open source is a sociological problem more than a technical one.
Papers I keep coming back to
- Attention Is All You Need — Vaswani et al., 2017. Annual re-read.
- Sparse Autoencoders Find Highly Interpretable Features — Bricken et al., Anthropic.
- In-Context Learning and Induction Heads — Olsson et al., Anthropic.
Got a recommendation? Email me — I read every one.