Vincent Sitzmann

Vincent Sitzmann is an Assistant Professor in the MIT Department of Electrical Engineering and Computer Science and a principal investigator at CSAIL, where he leads the Scene Representation Group. His research investigates how neural scene representations, implicit neural fields, and generative models enable machines to perceive, model, and act within the physical world with greater autonomy and generalization. Sitzmann’s work bridges computer vision, machine learning, and robotics, exploring topics such as 3D reconstruction, view synthesis, diffusion models, geometric reasoning, world modeling, and the design of learned representations that support robust control. Prior to joining MIT, he received his PhD from Stanford University, where he contributed foundational work on neural radiance fields and implicit representations. His publications have appeared in CVPR, NeurIPS, ICLR, SIGGRAPH, and Nature, including award-winning papers on generalizable reconstruction and view synthesis. The Scene Representation Group focuses on building intelligent systems that understand complex environments through rich, interpretable models, with applications spanning robot control, generative simulation, and future embodied AI systems.