Research

My work spans generative molecular design, virtual cell modeling, and agentic systems for scientific discovery. For a complete list of publications, see my Google Scholar profile.

Projects

Virtual Cells

AI models that predict and explain how cells respond to chemical and genetic perturbations, to accelerate the discovery of new therapies. This is the team I currently lead at Valence Labs.

Read more on our Substack
LOWE preview
LOWE

An LLM-orchestrated workflow engine for drug discovery and one of the earliest agentic AI systems in the space. LOWE composes ML models, lab automation, and proprietary data to help researchers formulate, test, and prioritize hypotheses at scale.

SAFE

A molecular representation framework that enables a single generative model to handle all major molecular design tasks: de novo generation, scaffold decoration, and linker design.

Watch the talk
datamol.io preview
datamol.io

An open-source ecosystem of Python packages for molecular processing, including datamol (molecular manipulation), molfeat (molecular featurization), and safe (SAFE molecular design).

Selected Publications