Mentoring
I co-mentor student teams in UC Berkeley’s Data Science Discovery program, part of the College of Computing, Data Science, and Society, on AI/ML projects relevant to drug discovery. Together with Merck colleagues, I’ve mentored close to 20 students across five projects over three semesters (most recently Spring 2026), an ongoing collaboration each term. Two representative projects are below.
Agents in the Loop for Small Molecule Drug Design
AI/MLDrug Discovery
Student team: Akansha Jain, Ava Joshi, Annie Gao, Gautam Ramkumar, and Shreyash Goli, mentored by Alan Cheng, Claire Suen, David He, Xiaodong Zhu, and me (Merck & Co.). The project explores agentic workflows for small-molecule drug design.
This poster won a Research and Storytelling Award (also recognized as best poster) at the CDSS Data Discovery spring symposium, Spring 2026. Credit for the award belongs to the student team and the full five-person mentoring group.
Poster: cdss.berkeley.edu
LLM-Orchestrated Mixture-of-Experts for Solubility Prediction
AI/MLCheminformatics
Student team: Samantha Alonso, Surya Appana, Ameya Kiwalkar, and Hilary Wang, mentored by Alan Cheng, Claire Suen, David He, Xiaodong Zhu, and me (Merck & Co.). The project explores an LLM-orchestrated mixture-of-experts approach to molecular solubility prediction.
Poster: cdss.berkeley.edu
Other Research I’ve Supported
Attribution-Guided Genetic Operators for Drug Optimization
AI/MLDrug Discovery
I informally advised a post-baccalaureate researcher at Merck on this generative AI project: using attribution scores to guide where a genetic algorithm modifies a molecule during drug optimization. The paper was accepted as an oral presentation at the ICML 2026 AI for Science workshop. I’m thanked in the acknowledgments for helpful discussions. I am not an author on this work.
| Download paper here | OpenReview |
