Mihir Bafna

I am a first-year Ph.D. student in EECS at MIT advised by Bonnie Berger, interested in deep learning for structural biology and drug discovery. I am also a Machine Learning Scientist at Liquid AI, where I work on PLMs.

Previously, I completed my B.S. in CS from Georgia Tech specializing in the machine learning and theory threads. There, I worked on graph representation learning for spatial transcriptomics and was incredibly fortunate to have been advised by Xiuwei Zhang (GT), Jian Ma (CMU), and Ruochi Zhang (Broad).





selected papers


  1. biorxiv
    Generating functional and multistate proteins with a multimodal diffusion transformer
    Bowen Jing, Anna Sappington,  Mihir Bafna, Ravi Shah, Adrina Tang, Rohith Krishna, Adam Klivans, Daniel J Diaz, and Bonnie Berger
    bioRxiv Sep 2025
  2. arxiv
    Learning residue level protein dynamics with multiscale Gaussians
    Mihir BafnaBowen Jing, and Bonnie Berger
    arxiv Sep 2025
  3. PNAS
    Sparse autoencoders uncover biologically interpretable features in protein language model representations
    Onkar GujralMihir BafnaEric Alm, and Bonnie Berger
    Proceedings of the National Academy of Sciences Sep 2025
  4. biorxiv
    DANGO: Predicting Higher-Order Genetic Interactions
    Ruochi ZhangMihir BafnaJianzhu Ma, and Jian Ma
    biorxiv Jan 2025
  5. MLSB
    DiffRNAFold: Generating RNA Tertiary Structures with Latent Space Diffusion
    Mihir Bafna, Vikranth Keerthipati, Subhash Kanaparthi, and Ruochi Zhang
    NeurIPS Workshop on Machine Learning for Structural Biology Dec 2023
  6. ISMB
    CLARIFY: Cell-cell interaction and gene regulatory network refinement from spatially resolved transcriptomics
    Mihir BafnaHechen Li, and Xiuwei Zhang
    ISMB and Bioinformatics Jun 2023