Jiajun Wu

Jiajun Wu is an Assistant Professor of Computer Science and, by courtesy, of Psychology at Stanford University. He received his PhD and SM in Electrical Engineering and Computer Science from the Massachusetts Institute of Technology, advised by Bill Freeman and Josh Tenenbaum, and completed his undergraduate studies at Tsinghua University. Before joining Stanford in 2020, he was a Visiting Faculty Researcher at Google Research, working with Noah Snavely. Wu's research centers on computer vision, machine learning, robotics, and computational cognitive science, with a focus on physical scene understanding. His group develops AI systems that perceive, reason about, and interact with the physical world, drawing inspiration from human cognition and the natural environment. Key research areas include multi-modal perception from visual, acoustic, and tactile signals, such as the ObjectFolder and RealImpact datasets; visual generation of 4D physical scenes, exemplified by projects like 3D-GAN, pi-GAN, Point-Voxel Diffusion, and WonderWorld; neuro-symbolic visual reasoning with physical concept grounding, including NS-VQA, Shape Programs, CLEVRER, and LEFT; and robotics and embodied AI, such as RoboCook and BEHAVIOR benchmarks. He is affiliated with the Stanford Vision and Learning Lab, Stanford AI Lab, Stanford HAI, Bio-X, and Wu Tsai Institutes.

Wu has garnered major awards and honors, including the NSF CAREER Award (2024), ONR Young Investigator Program (2024), AFOSR Young Investigator Program (2023), Okawa Research Grant (2024), IEEE Intelligent Systems AI's 10 to Watch (2024), MIT Technology Review Innovators Under 35 Asia Pacific (2024), Samsung AI Researcher of the Year (2020), ACM Doctoral Dissertation Award Honorable Mention (2019), AAAI/ACM SIGAI Doctoral Dissertation Award (2019), and MIT George M. Sprowls PhD Thesis Award (2020). He has received best paper awards and finalists at conferences including CVPR, ICCV, SIGGRAPH Asia, ICRA, CoRL, and IROS, along with faculty research awards from Google, J.P. Morgan, Meta, Amazon, and Samsung. Notable publications include "Learning a Probabilistic Latent Space of Object Shapes via 3D Generative-Adversarial Modeling" (NeurIPS 2016), "Visual Dynamics: Probabilistic Future Generation via Layered Cross Convolutional Networks" (TPAMI 2019), "The Neuro-Symbolic Concept Learner: Interpreting Scenes, Words, and Sentences from Natural Supervision" (ICLR 2019), "3D Shape Generation and Completion through Point-Voxel Diffusion" (ICCV 2021), "Birth and Death of a Rose" (CVPR 2025), and "The Scene Language: Representing Scenes with Programs, Words, and Embeddings" (CVPR 2025). With nearly 49,000 citations on Google Scholar, his work has profoundly impacted AI, computer vision, and robotics. Wu teaches "Minds and Machines" and supervises independent research projects.

Professional Email: jiajunwu@cs.stanford.edu