Technical Lead, Generative AI at Snap Inc
Los Angeles, CA, United States
Michael Vasilkovsky
Council Member
Michael Vasilkovsky is a technical leader specializing in generative AI and neural rendering, with a focus on 3D technologies that power augmented reality experiences. Since joining Snap Inc. in 2021, he has led the development of foundational 3D models and their extensions, delivering business impact through cutting-edge advancements in generative AI.
As a Technical Lead on Snap's Generative AI team, Michael spearheaded the development of several flagship products within the GenAI Suite, including AI Video Lenses powered by in-house video models, 3D asset generation tools that outperform solutions from industry competitors, and HeadMorph—the first-in-market generative 3D avatars driven by face tracking. These tools have driven over billions impressions worldwide, becoming foundational for Snap's AR ecosystem which serves 450 million daily active users.
His research contributions include work on video generation models with camera control (VD3D), 3D reconstruction techniques, and spatially aware multiview diffusers. As a published researcher, Michael's work has appeared in prestigious venues including CVPR, SIGGRAPH, ICLR, and AAAI, establishing him as a recognized authority in neural rendering and generative AI for 3D applications.
As a Technical Lead on Snap's Generative AI team, Michael spearheaded the development of several flagship products within the GenAI Suite, including AI Video Lenses powered by in-house video models, 3D asset generation tools that outperform solutions from industry competitors, and HeadMorph—the first-in-market generative 3D avatars driven by face tracking. These tools have driven over billions impressions worldwide, becoming foundational for Snap's AR ecosystem which serves 450 million daily active users.
His research contributions include work on video generation models with camera control (VD3D), 3D reconstruction techniques, and spatially aware multiview diffusers. As a published researcher, Michael's work has appeared in prestigious venues including CVPR, SIGGRAPH, ICLR, and AAAI, establishing him as a recognized authority in neural rendering and generative AI for 3D applications.
Prior to Snap, Michael led AI projects at Neu.ro, designing real-time pose estimation systems and super-resolution technologies for major telecom vendors. He holds a Master's degree in Applied Mathematics and Computer Science from Skolkovo Institute of Science and Technology and studied at the Yandex School of Data Analysis and Moscow Institute of Physics and Technology.
Michael combines academic research with practical engineering expertise to develop innovative technologies that reshape how people create and interact with AR content. His work continues to advance the state of the art in video generation, focusing on improving generation speed and video length while maintaining quality, and developing more intuitive interfaces for camera control and content editing.
Michael combines academic research with practical engineering expertise to develop innovative technologies that reshape how people create and interact with AR content. His work continues to advance the state of the art in video generation, focusing on improving generation speed and video length while maintaining quality, and developing more intuitive interfaces for camera control and content editing.