I am a Master of Computer Science student at the University of Illinois Urbana-Champaign and am currently conducting robot learning research with the MSC Lab at UC Berkeley.
I am seeking Fall 2027 Ph.D. opportunities and industry opportunities in robotics. I welcome conversations about research, collaborations, and other opportunities—please feel free to reach out.
Prior to UIUC, I was admitted to the PhD program in Finance at the University of Rochester, where I later earned a master’s degree. Deciding to master out and restart on a new path was one of the most difficult but also one of the most important decisions in my academic journey. My training at Rochester provided me with a strong foundation in research, quantitative reasoning, and principled problem formulation. Over time, I realized that my core interests lie in AI and robotics, with the goal of building intelligent systems that can perceive, reason, and act in the real world. This experience not only shaped my research perspective, but also gave me the clarity and conviction to pursue the problems that matter most to me.
I received my bachelor’s degree in Financial Technology, with a second major in Computer Science, from the University of Hong Kong, where I built a strong foundation in quantitative analysis and computing, and developed an early interest in machine learning through coursework.
This project learns a shared interaction representation from synthetic robot trajectories, human hand motion, and human-anchored robot trajectories. It supports zero-shot generalization to new objects and few-shot adaptation to new embodiments and tasks with frozen encoders.
ChildPose shows that pediatric pose estimation needs child-specific temporal and developmental modeling beyond fine-tuning adult-centric estimators. It combines memory-augmented temporal attention with age-aware co-training to produce stable video predictions, outperforming off-the-shelf and child-data fine-tuned baselines on unseen subjects.
V2AP converts object-centric priors from human videos into robot-usable grasp supervision. It verifies candidate grasps in simulation, then aggregates successful rollouts into affordance posteriors that improve grasp success over raw human priors.
ProCLIP combines complementary representation spaces to capture diverse multimodal semantics. With pretrained backbones fixed, it improves image–text retrieval over single-geometry baselines.
Read the paper