People

Yilun Du is an Assistant Professor at Harvard in the Kempner Institute and Computer Science. He received his PhD from MIT EECS, advised by Leslie Kaelbling, Tomas Lozano-Perez, and Joshua B. Tenenbaum. He also holds a bachelor's degree from MIT, was a research fellow at OpenAI and a senior research scientist at Google DeepMind.

His research focuses on developing intelligent embodied agents in the physical world through generative AI, decision making, and robot learning.

Ruojin Cai is a Postdoctoral Research Fellow at the Kempner Institute for the Study of Natural and Artificial Intelligence at Harvard University, working with Professor Yilun Du. She received her Ph.D. in Computer Science from Cornell University, co-advised by Professors Bharath Hariharan and Noah Snavely, and holds a B.S. in Computer Science from Tsinghua University.

Ruojin's research focuses on 3D computer vision and spatial intelligence, developing generative and multimodal foundation models to enhance 3D reconstruction and understanding.

Haonan Chen is a postdoc at Harvard, advised by Yilun Du. Previously, he was a Ph.D. candidate in Electrical and Computer Engineering at the University of Illinois Urbana-Champaign, advised by Katie Driggs-Campbell and worked closely with Yunzhu Li.

He designs robots that can see, reason, and act in complex environments through multimodal decision making. His research develops composable models that integrate visual, spatial, tactile, and temporal cues to predict outcomes and plan robust actions. His work has been recognized with a Best Paper Finalist at CoRL 2023 and a Best Paper Award at the ICRA 2025 Workshop on Foundation Models and NeSy AI.

Weirui Ye is a Postdoc Fellow in LIS Group at MIT CSAIL, working with Prof. Leslie Pack Kaelbling and Tomás Lozano-Pérez at MIT and Yilun Du at Harvard. Previously, he received his Ph.D. from Tsinghua University under the supervision of Prof. Yang Gao. During his Ph.D., he also worked as a visiting scholar at UC Berkeley advised by Prof. Pieter Abbeel.

His research goal is to develop human-level policy learning algorithms for robots in learning efficiency and generalization. He is interested in Reinforcement Learning, Generative World Models and Lifelong Learning for Robots.

Wenhui (Oscar) Huang is a postdoctoral researcher at Harvard University, co-advised by Yilun Du and Heng Yang. He received his Ph.D. from Nanyang Technological University and his Master's degree from Politecnico di Milano.

His research focuses on generative AI (GenAI), vision-language models (VLMs), and vision-language-action models (VLAs), with an emphasis on their applications in the physical world, particularly in robot learning and autonomous driving.

Anand Gopalakrishnan is a Postdoctoral Fellow in AI and Cognitive Science jointly advised by Yilun Du and Samuel Gershman at Harvard University. He received his PhD in Informatics from USI Lugano (Switzerland) advised by Jürgen Schmidhuber.

Anand's research is broadly centered around deep learning and representation learning with a focus on modularity and compositionality. Psychological studies, cognitive and neuroscientific theories on human perception like attentional gaze, saliency, object perception, binding, relational reasoning, categorization etc. inform his design of computational models. He is deeply fascinated by both natural and artificial intelligence and wishes to distill the essential computational principles from the human mind to build more human-like AI.

Zhenting Qi is a CS PhD student at Harvard University, co-advised by Yilun Du and Hima Lakkaraju. He previously received his Master's degree at Harvard.

His research is centered on reasoning and reliability in AI systems, with a recent emphasis on developing agents that integrate code development and scientific discovery.

Haldun Balim was born in Turkey, where he studied computer science at Koç University. He then completed his M.Sc. in Robotics, Systems, and Control at ETH Zurich, focusing on the intersection of control theory and machine learning.

His research lies at the intersection of control theory and machine learning, aiming to design data-driven and provably correct solutions with a focus on safety, reliability, and verification in uncertain, human-centric environments. He seeks to develop holistic, principled methods that unify tools from control, machine learning, optimization, and perception to enable efficient and trustworthy decision-making.

Zhiyi Li is a PhD student at MIT EECS advised by Yilun Du. He graduated from Tsinghua University with a major in electronic engineering.

He is interested in exploring generative modeling for embodied agents to understand and interact with the world.

Sarah Liaw is a first-year PhD student working with Profs. Yilun Du and David Alvarez-Melis.

She is interested in developing models that uncover the physical and structural principles underlying data, enabling generalizable learning in robotics and scientific domains.

Haoyu Xiong is a PhD student at MIT EECS, co-advised by Prof. Antonio Torralba and Prof. Yilun Du.

He is interested in Robot Learning.

Amani Kiruga is a PhD student at Harvard University advised by Yilun Du. He previously worked on 3D vision and generative modeling with Ayush Tewari and Vincent Sitzmann at MIT and received his B.S. in Computer Science from the University of Delaware in 2024.

His research focuses on generative models, 3D generalization, and their applications to embodied intelligence and robotics.

Hang Le is a first-year PhD student working with Prof. Yilun Du.

Hang Le is interested in building embodied agents that can learn to understand and interact with the physical world, and have sufficient sensorimotor skills to follow complex instructions and adapt to diverse environments.

Aayush Karan is a third-year Ph.D. student in Computer Science at Harvard, advised by Prof. Sitan Chen and Prof. Yilun Du. He graduated from Harvard University in 2023 with an AB in Physics and Mathematics and an SM in Computer Science. Prior to his Ph.D., Aayush conducted research in pure mathematics with Dr. Jianfeng Lin and Dr. Paul Terwilliger and developed RNA structure prediction algorithms with Dr. Elena Rivas.

Aayush's research broadly targets the foundations of generative models, focusing on delivering algorithmic insights that enhance intelligent capabilities such as reasoning, reward optimization, and failure prediction.

Hansen Lillemark is currently a second year PhD student at UC San Diego advised by Profs. Rose Yu and Taylor Berg-Kirkpatrick. He previously studied CS at UC Berkeley, advised by Prof. Bruno Olshausen and Kurt Keutzer, and in a past life, founded a financial technology startup.

Hansen's current research is focused on memory in world models: how can a generative model selectively compress previous observations and information into an opinionated world state, and use this memory for predicting the future? More broadly, he is interested in incorporating concepts from neuroscience such as grid and place cells into generative modeling.

Zoe Wu graduated from Harvard University with a degree in Statistics in 2025 and is currently a post-bacc at Kempner, advised by Yilun Du.

She is interested in understanding how the theoretical frameworks of differential geometry, statistical mechanics, and dynamical systems can be applied to improve generative AI systems.

Runqian Wang is an undergraduate student at MIT double majoring in AI and Math. He is currently a visiting researcher working with Yilun.

His focus is on generative modeling, particularly image and video generation. He is most interested in simple yet effective methods that can generalize.

Lillian Sun is an undergraduate student at Harvard studying CS and Statistics, along with a concurrent Masters in CS. Her research focuses on improving communication between models in multi-agent systems. She is interested in using synthetic data for self-improvement and designing incentives for robustness and alignment.

Research interests and focus areas.

Jiawei Gao is an undergraduate at Tsinghua University. Previously he worked at CMU RI.

He focuses on learning physics-structured world models for robotics and cognitive-science inspired robot learning.

Runhan Huang is a senior student at Tsinghua University, Yao Class. In the spring of 2025, he served as a research intern in Professor Du's laboratory. Prior to that, he worked as a research assistant at both the Shanghai Qizhi Institute and Tsinghua University.

His research interests lie at the intersection of robotics and generative AI. He aims to build generalizable embodied agents that enable robust and adaptive interaction and planning in complex real-world environments.

Xiaoshen Han is a final-year undergraduate student in the IEEE Honor Class at Shanghai Jiao Tong University, majoring in Computer Science. He is now a visiting undergraduate intern at Kempner Institute, advised by Prof. Yilun Du.

His research interest lies in reinforcement learning, robotics, and sim-to-real. He is passionate about designing novel algorithms that enable robots to tackle more challenging tasks and generalize across a broader range of tasks.

Chaoqi Liu is an undergraduate student at the University of Illinois Urbana-Champaign. He is currently a visiting student intern at the lab, where he works on research in robotics.

Chaoqi is broadly interested in robotic manipulation. His past research has included structured policy learning and dynamics/world model learning, and he is currently focusing on action representation learning for robot policies.

Peilin Wu is a senior visiting undergraduate from Shanghai Jiao Tong University, currently working with Professor Yilun on multimodal world models.

Peilin is broadly interested in robotics and machine learning. Specifically, she's focusing on constructing comprehensive world models for robot decision-making, which can integrate information from diverse modalities and learn from interactions with the real world. She also has experience working with quadruped robots and reinforcement learning.