I apply generative AI and physics-based simulation to design and optimize small-molecule therapeutics, combining cheminformatics, deep learning, and molecular simulation to optimize potency, selectivity, and ADMET properties during lead optimization’s design-make-test-analyze cycles.
Research Highlights
Generative AI & Molecular Design
AI/MLDrug Discovery
GFlowNet-based generative models for R-group optimization and core-hopping, constrained to validated reaction templates and commercially available building blocks.
Molecular Property Prediction
AI/MLCheminformatics
Automated benchmarking pipeline across multiple architectures for ADMET prediction. 25th / 100 finalists in OpenADMET Blind Challenge.
CheMeleon-Rxn: Reaction Property Foundation Model
AI/MLCheminformatics
Pre-trained on ~2M reactions, fine-tuned for reaction property prediction. Best-or-tied-best on 7 of 8 low-data benchmarks.
Mentoring & Student Research
AI/MLDrug Discovery
Co-mentor UC Berkeley Data Science Discovery teams on AI/ML for drug discovery. Research and Storytelling Award, CDSS Spring 2026 symposium.
Polyelectrolyte Simulations & Ion Binding
PhysicsSimulation
Enhanced-sampling MD revealing ion-ion correlations as the primary driver of polyelectrolyte attraction, with autoencoder analysis of complex structures.
| View all research areas | Full publication list |
Focus Areas
Generative AI Molecular Design De Novo Design Lead Optimization ADMET Prediction Graph Neural Networks Deep Learning Multi-Objective Optimization Drug Discovery Cheminformatics Property Prediction Computational Chemistry Structure-Activity Relationships Molecular Dynamics
Education
| Ph.D., Chemical Engineering | California Institute of Technology, 2024 |
| MS, Chemical Engineering | California Institute of Technology, 2022 |
| BS, Chemical Engineering | University of California, Berkeley, 2019 |