Research

AI agents, EHR foundation models, spatial and single-cell omics, and integrative genomics.

I develop AI agents and foundation models at the intersection of computational biology and bioinformatics, working across spatial omics, genetics, and clinical data.

01

AI Agents for Biological Discovery

How can biological interpretations from omics analysis be checked against supporting evidence?

Building agent harnesses for context-aware, evidence-grounded biological discovery.

I design agent harnesses that combine task-specific skills, MCP-connected tools and knowledge sources, custom guardrails, and bounded repair loops driven by validation. The workflows adapt to cell and tissue context, with biological interpretations checked against supporting evidence.

  • Task-specific skills
  • MCP
  • Custom guardrails
  • Validation & repair
Explore related software
An agent harness combines task-specific skills and MCP-connected resources within custom guardrails. Validation sends failed steps through a bounded repair loop and passes supported results to the output.
02

Foundation Models for Electronic Health Records

How can we learn shared clinical representations from notes and structured EHR data?

Building clinical foundation models by aligning clinical notes and structured EHR data through contrastive learning.

Clinical notes provide narrative context, while structured EHR data capture clinical measurements and events. I use contrastive learning to connect these complementary views and learn shared clinical representations.

  • Foundation models
  • Contrastive learning
  • Clinical notes
  • Structured EHR
Clinical notes and structured EHR data feed a contrastive learning process that aligns their representations in a shared clinical space.
03

Spatial & Single-cell Omics

How do cell states and genetic effects vary across tissue regions?

Developing computational tools for spatial omics and single-cell eQTL data.

I am interested in methods that connect molecular measurements to cell states, tissue context, and regulatory variation, building on my experience with functional genomic data.

  • Spatial omics
  • Single-cell analysis
  • Functional genomics
Related publications Integrative immunogenomics
Spatially arranged cells in a tissue linked to a gene-expression matrix to connect cell states with tissue context.
04

Integrative Genomics

Which genes and molecular processes link genetic variation to complex disease?

Combining GWAS, QTL, transcriptomic, proteomic, and multi-omics data to study disease mechanisms.

I bring together doctoral training and postdoctoral work on TWAS, GWAS, WES, and multi-omics association studies to study complex traits and diseases.

  • TWAS and GWAS
  • Multi-omics
  • Complex traits
Related publications Scalable statistical genetics Integrative multi-omics GWAS method development
GWAS, RNA, and protein signals integrated into a network connecting genes and disease biology.

Several of these directions are supported by reusable software and analysis tools. See the software page for current projects.