Research Interests

Graph Neural Networks

We explore the frontier of graph-based learning, where the structure of data is as vital as the data itself. Our work in Graph Neural Networks (GNNs) seek to model complex relational systems, such as social networks, urban networks, and biological processes, by learning from both topology and feature interactions. We focus on both theoretical advancements (e.g., expressivity, heterophily-aware models, scalability) and high-impact applications in domains where connectivity is not optional, but fundamental. 

Our research pushes GNNs beyond academic prototypes, enabling them to power real-world decision-making systems with transparency and robustness.

Human Computer Interaction & Data Visualization

At the intersection of human cognition and computational systems, our Human Computer Interaction (HCI) and Visualization research is grounded in singular belief: data must be seen to be understood, and felt to be believed. We design intuitive, expressive, and aesthetically powerful interfaces that transform abstract data into compelling narratives and actionable insights. 

Whether it’s through interactive dashboards, visual analytics systems, or exploratory environments for large-scale data, our goal lies in amplifying human intelligence and bridging the gap between algorithmic output and human understanding. 

Empirical Causal Inference

In a world awash with correlations, we ultimately seek cause. Our empirical causal inference research harnesses both traditional statistical theory and modern machine learning to identify, estimate, and validate causal effects from observational data. We are especially interested in high-dimensional, real-world settings, such as policy evaluation, market dynamics under uncertainty, and social interventions, where rigor meets uncertainty. From synthetic control methods to instrumental variables and beyond, we work at the cutting edge of methodology to inform decisions that matter. 

Our mission is clear: to move from what is to what would have been, and ultimately to what should be.