Why I'm writing a weekly paper review

I read a lot of papers. Between running an AI and bioinformatics team and trying to keep up with a field that reinvents itself every few months, reading closely is part of the job — and honestly, one of the best parts. This is where I'll share that reading, one paper a week.

What this is

Each week I'll take a single paper that's shaping AI-driven drug discovery and write a short, opinionated review. Not a summary you could get from the abstract — a working scientist's take: what's genuinely new, what the results actually support, and where I'd be cautious before betting a program on it.

What I'll cover

• Generative models for antibodies and other biologics
• Protein structure prediction and de novo design
• Molecular machine learning and property prediction
• Benchmarks, data, and the gap between a leaderboard and the lab

The questions I bring to every paper

Three, mostly. Does the method beat a fair baseline, or a convenient one? Would the result survive contact with real discovery data and its noise? And if it works, what does it let us do next week that we couldn't do last week? If a paper answers those well, I'll say so. If it doesn't, I'll say that too.

New posts land weekly. If something here sparks a thought — or you think I've read a paper wrong — tell me.