James Hu

James Hu
Building machine learning methods for precision medicine and biology.

I’m a recent graduate of the University of Florida, where I majored in computer science with a minor in mathematics. I’m interested in how we can improve AI to understand biological systems and the origins of complex disease.

Recently, I worked with Lorin Crawford in the Biomedical Machine Learning group at Microsoft Research New England on Project Ex Vivo, a joint effort with the Broad Institute. My work explored lab-in-the-loop experiment acquisition frameworks for improving the out-of-distribution generalization of single-cell foundation models.

Previously, I researched interpretable neural networks at the Broad Institute and deep learning for risk stratification at Massachusetts General Hospital.

Research Interests

  • Emergent Complexity in Biological Systems

    How simple biological components organize and interact to give rise to complex, system-level behavior.

  • Multimodal Representation Learning

    Learning shared representations across different biological measurements.

  • Understanding Complex Disease

    Decoding the etiologies behind complex, polygenic diseases.

  • Foundation Models for Single-Cell Biology

    Models that learn general-purpose representations of single-cell biology and keep improving through continual learning and lab-in-the-loop experimentation.

  • Mechanistically Interpretable AI

    Biologically informed models whose internal mechanisms can help explain their predictions.

Experience & Education

  • —

    Microsoft Research New England

    Research Intern · Biomedical Machine Learning

    Mentored by Lorin Crawford on Project Ex Vivo, a joint research effort with the Broad Institute. Lab-in-the-loop experiment acquisition for improving the out-of-distribution generalization of single-cell foundation models.

  • —

    University of Florida

    Teaching Assistant
  • —

    Broad Institute of MIT and Harvard

    Computational Research Intern · Broad Summer Research Program

    Interpretable neural network approaches for optimizing pathway-specific polygenic risk scores.

  • —

    Massachusetts General Hospital

    Machine Learning Research Intern

    Deep learning-based algorithms for data-driven risk stratification.

Education · August 2022 — May 2026

University of Florida

Bachelor of Science in Computer Science · Minor in Mathematics · 3.98 GPA

UF Honors Program · University Research Scholars Program · Benacquisto Scholarship

Contact

Feel free to reach out by email, or find me on GitHub, LinkedIn, and X.