research

AI for Protein Structure Prediction

My current work studies how to combine AlphaFold-style structure priors with limited cryo-EM observations. The central goal is to preserve the strong structural prior learned by modern protein-structure models while allowing target-specific correction from experimental particle data.

Computational Imaging

I am interested in inverse problems and differentiable forward models for scientific imaging. In cryo-EM, this includes connecting atomic coordinates to particle-level observations through density generation, projection, and experimental supervision.

Efficient Scientific Machine Learning

My earlier work studied machine learning for terahertz spectroscopy, with an emphasis on compact, efficient, and robust models for molecular classification.