Computational Biology Portfolio
San Francisco, USA
Computational biologist and bioinformatician at UCSF with 4+ years of hands-on work in single-cell, single-nuclei, and spatial transcriptomics. I build production-grade analysis pipelines and research software for liver disease and tumor microenvironment discovery. I'm also developing a foundation model for the liver.
Production tools and modeling systems for transcriptomics and computational pathology.
Foundation Model
Liver-specific transformer (23.6M parameters) pre-trained using masked gene prediction on 1.04M human liver cells (36K genes, 7 diseases, 125 cell types), with strong transfer to spatial transcriptomics tasks including hepatocyte zonation.
Spatial Genomics
Python package for inside/interface/outside tumor region assignment in spatial transcriptomics using nearest-neighbor context, with built-in visual diagnostics and downstream expression profiling.
Spatial + Atlas
Toolkit for bridging spatial transcriptomics outputs with CellxGene-style exploratory workflows, including utilities for region-aware expression analysis and dataset handoff for interactive atlas inspection.
Julia Toolkit
Open-source Julia package for transcriptomics analysis from preprocessing through clustering, marker discovery, and UMAP, built as a high-performance alternative to typical R/Python workflows.
AI Scientist
End-to-end Python framework that combines single-cell tumor profiles with LLM reasoning to classify hepatoblastoma subtypes and generate biologically interpretable hypotheses with supporting literature context.
Syncing live metadata for public repositories...
Current focus areas and selected publications.