Research

Research Assistant — Multi-Modal Single-Cell Deep Learning

University of Michigan

I am currently a research assistant in the Liu Lab at the University of Michigan, working on multi-modal deep learning methods for single-cell biological data, with a focus on integrating heterogeneous modalities such as RNA-seq and ATAC-seq. My work explores computational approaches for learning shared representations across modalities and improving downstream biological prediction tasks.

In this project, I work with PyTorch-based research codebases, reproducing models from recent literature and extending existing architectures by modifying model components and training pipelines. Through systematic experimentation, I evaluate model performance on downstream tasks and analyze how architectural and training choices affect cross-modal representation learning.

This experience has strengthened my understanding of multi-modal machine learning, research-oriented software development, and data-driven problem solving in computational biology. It has also exposed me to the full research workflow, including literature review, experimental design, and iterative model refinement.