I am a PhD student at Columbia University in the department of Electrical Engineering and Zuckerman Mind Brain Behavior Institute, working with Prof. Nima Mesgarani. Prior to Columbia, I earned my master’s degree in ECE from University of Michigan and bachelor’s degree in Computer Science from Fudan University. I was fortunate to have worked with Prof. Yuanning Li at ShanghaiTech, Prof. Jonathan Brennan at U-M and Dr. Mark Richardson at Harvard Medical School.

Research

My research sits at the intersection of machine learning, neuroscience, and linguistics. For the first time in human history, we possess a non-human intelligence that approaches human-level capabilities. While both the human brain and contemporary AI systems remain “black boxes,” carefully designed experiments can harness one to illuminate the other. My research aims to establish a bidirectional link between artificial and neurobiological intelligence systems. I focus on multimodal approaches that combine computational modeling with invasive neural recordings, seeking to better understand the brain and ultimately improve patient care. Specifically, my work is organized around three main directions:

• Interpreting and Designing AI Models
  My first line of research aims to “open the black box” of machine intelligence by interpreting existing AI models and developing new, more interpretable ones. By dissecting how AI systems represent and process information—especially in the realm of language—we can highlight parallels (and differences) with human cognitive function. This line tries to contributes to the field of AI interpretability but also lays the methodological foundation for using these models in neuroscience research.

• Uncovering the Neural Mechanisms of Human Cognition through AI Models
  Building on the progress in interpretable AI, my second research direction uses these models as a lens through which we can “open the black box” of the human brain. By aligning interpretable AI models with neural data—whether it involves language, speech, or other domains like music, code, or mathematics—we can systematically probe the representations and transformations that underlie human cognition.

• Translational Studies for Brain-Computer Interfaces and Assistive Technologies
  Finally, a natural extension of these inquiries lies in clinical and translational applications. By leveraging interpretable AI models to decode speech and language from invasive electrophysiology, we aim to develop advanced brain-computer interfaces (BCIs) that can restore communication to individuals with severe speech or motor impairments.

Through these three directions, I aim to leverage AI advances to illuminate both machine intelligence and the human brain, fostering progress in theory and application for scientific and societal benefit.

Reach out to me!

I’ve always enjoyed talking about science and exploring new ideas. If you’re interested in collaborating or simply having a conversation about science, feel free to reach out!