About Me
I am a first-year Ph.D student in the Department of Computer Science at Virginia Tech, fortunately advised by Dr.Xuan Wang. I am also a student member of the Sanghani Center for Artificial Intelligence and Data Analytics at VT. Before that, I received my B.S. in computer science from Wuhan University.
Research
My research studies scalable compositional intelligence: building AI systems by composing foundation models rather than scaling monolithic ones, leveraging modularity for stronger generalization and sample efficiency. I approach this from three connected perspectives: composable building blocks, integration mechanisms that combine them, and systems that scale composition across them.
❶ Representation — Foundation Models
How can foundation models be made composable and reusable as building blocks?
I develop foundation models for diverse modalities, including text, graphs, and biological sequences (e.g., proteins, DNA), focusing on representations and interfaces that make each model modular and reusable as a building block for composition.
Representative work: BioArc, UniGLM.
❷ Integration — Multimodal Models and Agents
How should foundation models be combined to solve a given task?
I study integration mechanisms that combine heterogeneous foundation models, ranging from learned multimodal fusion to agent frameworks that select, invoke, and combine FMs at inference time, enabling adaptive and goal-directed composition.
Representative work: GraphGPT-O, GAugLLM.
❸ Interaction — Systems
How can composition scale efficiently across multiple agents and models working together?
I design multi-agent communication and coordination mechanisms that enable models and agents to collaborate effectively, supporting efficient and scalable composition in large-scale systems.
Together, these three perspectives form my long-term vision: to understand how complex intelligence can emerge from the efficient composition of simpler foundation models.
Service
- Reviewer: KDD 2025, ICML 2026, ARR 2025
- Student Volunteer: KDD 2024