About

I am an incoming Ph.D. student in Electrical and Computer Engineering at the University of Michigan, Ann Arbor, where I will be advised by Prof. Ziyou Song.

Before joining UMich, I am completing my B.S. in Electrical Engineering at Tsinghua University. My research interests lie broadly in energy storage systems, battery management systems, sustainable transportation, and grid integration. I am particularly interested in physics-based modeling, state estimation, optimal control, and data-driven methods for reliable and efficient energy systems.

Beyond energy systems, I have also worked on AI agents for software debugging and neural architecture search for biological foundation models. These experiences have shaped my broader interest in building reliable, data-driven, and intelligent engineering systems.

Research Interests

Energy Storage Systems: physics-based modeling, SOC/SOH estimation, and optimal control of batteries.
Sustainable Transportation: battery management systems and thermal management for electrified vehicles.
Grid Integration: using energy storage to bridge transportation systems and renewable power grids.
AI for Scientific and Engineering Systems: LLM agents, neural architecture search, and data-driven modeling.
AI for Engineering Systems: LLM agents, neural architecture search, and data-driven modeling.

Publications

BioArc: Discovering Optimal Neural Architectures for Biological Foundation Models

Y. Fang, H. Xu, J. Han, S. Ding, Y. Wang, Y. Wang, and X. Wang

International Conference on Machine Learning (ICML), 2026.

PDF Manuscript

Selected Projects

Concurrency Debugging AI Agent

I am working on an LLM-orchestrated debugging agent for concurrency issues in distributed systems, including races, deadlocks, and ordering violations. This project focuses on hierarchical bug localization, repair suggestion generation, and action-level safety guardrails for autonomous debugging agents.

I am grateful to Prof. Tianyi Zhang and Prof. Yongle Zhang at Purdue University for their guidance and support on this project.

Neural Architecture Search for Single-Cell Foundation Models

I worked on neural architecture search for biological foundation models on single-cell sequencing data. This project explores diverse neural architectures, including CNN, LSTM, Mamba, Hyena, and Transformer-based models, across multiple biological tasks.

I am grateful to Prof. Xuan Wang at Virginia Tech for her guidance and support on this project.

CV

My full CV is available here.