Hi, I’m Nathan! I lead drug discovery and development research at Anthropic. Before that, I co-founded Coefficient Bio, where we figured out how to make biotech company creation and operation hill-climbable.
I’m interested in creating and using powerful AI systems to help scientists develop new medicines to cure disease, slow down or reverse aging, and improve the human condition. My team is building “pipeline-in-a-person” to enable a single skilled biotech operator + a bunch of Claudes to create and operate drug programs. We’re focused on making Claude the best model for biotech.
I was a group leader at Genentech, where I worked on lab-in-the-loop for therapeutic antibody design, discrete diffusion, interpretability, and foundation models for drug discovery. Before that, I did a PhD in Materials Science & Engineering at the University of Pennsylvania as an NDSEG fellow, and was a postdoctoral fellow at MIT. I occasionally write on the internet.
Writing
- Ask not why more talented people don’t work in biotech Dec 2025
- Legacy vs AI-native companies isn’t a fair fight Oct 2025
- Publishing and communicating research in AI/ML is fundamentally broken Jul 2025
- Getting started in BioML research & engineering May 2025
- The impact of AI/ML in drug discovery isn’t where you think it is Aug 2024
Research
Full list on Google Scholar. Selected work:
- Lab-in-the-loop therapeutic antibody design with deep learning bioRxiv · 2025
- Generalists vs. Specialists: Evaluating LLMs on highly-constrained biophysical sequence optimization tasks ICML 2025
- Protein discovery with discrete walk-jump sampling ICLR 2024 · Outstanding Paper
- Protein design with guided discrete diffusion NeurIPS 2023
- Neural scaling of deep chemical models Nat Mach Intell · 2023
Elsewhere
Get in touch
hey@ncfrey.com — if you’re reaching out cold, please read Sriram Krishnan’s guide to cold email first.