Scientific machine learning for biological discovery and operational decisions.
I am a molecular biologist and data scientist who builds systems that turn large search spaces into testable experiments and practical actions. I lead data science at UCB, publish first-author applied machine-learning research in protein engineering and multi-omics, and have worked across Pfizer, NetApp, and Lawrence Livermore National Laboratory. My background includes a PhD in molecular biology, an MS in data science, and a BS in biochemistry.
The thread connecting my work is simple: combine domain evidence with machine learning to narrow complex search spaces into decisions that can be tested.
- FOX gene candidate prediction: multi-omic machine learning and comparative bioinformatics for prioritizing genes involved in oxic nitrogen fixation. Paper | Interactive app | App source
- Active learning for Rubisco engineering: protein language model embeddings and sequential acquisition for prioritizing enzyme variants. Paper
- Nitrosomes: protein language modeling and live-cell imaging of condensate-like nitrogenase organization in heterocysts. Preprint
- Secondary metabolites and diazotrophy: cheminformatic ranking of cyanobacterial metabolites and likely diazotrophic producers. Paper
- deck-builder: structured, LLM-assisted generation of editable PowerPoint presentations.
- Insight Harness: a governed analytics workbench that keeps metric computation deterministic while using language models only for bounded translation and narration.
- AI2Analytics: reusable analytical pipelines with AI-assisted data discovery, configuration, and adapter generation.