About

My name is Nic Fishman, I'm currently a PhD student in Statistics at Harvard University.

I build reliable large-scale machine learning systems for science, focusing on biology, healthcare, and social science.

Currently I am working on three major applied projects:

1. Foundation models for causal perturbation prediction in single-cell genomics.

2. Dynamic treatment effects and policy learning from medical foundation models trained on electronic health records.

3. Non-convex optimization for estimating vote choice in elections, with applications in causal inference.

In addition to application-driven work, I also focus on machine learning and causal inference methodology, critiquing existing approaches and developing new methods for practitioners.

You can find a more complete CV here, and you can reach me at njwfish [at] gmail.com.