What This Site Is For

This site is my home base for work that sits between research, production AI systems, drug discovery, and teaching.

The near-term plan is simple: publish notes that make complicated systems easier to reason about. Some posts will unpack ideas from Build AI Drug Discovery Pipelines. Others will come from teaching graduate cheminformatics and machine learning at UC Berkeley, or from the practical messiness of building agentic systems that have to work outside a benchmark.

I want the writing here to be useful to people who build things: researchers, students, applied scientists, and drug discovery folks trying to make sense of modern ML without losing the thread of the underlying science.

This site is also the canonical archive. I may syndicate posts to Substack, LinkedIn, or X, but the version here is the one I will keep current. You can follow via RSS if you prefer the old, sturdy internet.

For now, the best starting points are the book, the talks page, and the publications list.