Research
I apply machine learning to functional genomics, with a focus on post-transcriptional regulation. Most of my work asks a single question in different forms: when a measurement says an RNA-binding protein binds somewhere in the cell, what is actually being measured?
My doctoral work improved the prediction of cellular binding sites of RNA-binding proteins by integrating in vivo and in vitro data through a recalibration framework. Alongside it, I collaborate with wet-lab groups on the role of RNA-binding proteins in acute myeloid leukemia, and I am interested in genomic foundation models and in agents that can carry out scientific work end to end.
Research Areas
RNA-Binding Proteins
Post-transcriptional regulation and how RNA-binding proteins choose their targets in cells.
Machine Learning for Genomics
Protein language models and genomic sequence-to-function models applied to functional genomics.
Translational Oncology
Translational control of leukemic stem-cell programs and driver-module discovery in cancer.
AI Agents for Science
Agentic pipelines that plan, call scientific tools, and synthesize evidence for biological questions.
Selected Projects
Recalibrating in vitro models for in vivo RBP binding
An RBP-agnostic CLIP model and an intrinsic-specificity score that flag when eCLIP binding reflects condensate localization rather than sequence specificity. Presented at ISMB/ECCB 2025 and CSHL AI in Biology 2026.
Translational control in acute myeloid leukemia
Quantifying translation efficiency from RiboSTAMP editing data to show that MSI2 promotes translation of a leukemic stem-cell program through a cooperative RBP module.
Protein representations for RBP specificity
Benchmarking protein language model (ESM) and AlphaFold-derived representations for predicting RBP binding specificity directly from amino-acid sequence.
Cancer driver modules from multi-omics data
DriveWays and MEXCOwalk, graph-based methods for identifying overlapping cancer driver modules from TCGA data.