Senior AI Research Scientist Manager at Meta, leading frontier generative media systems from research to production.
I specialize in large-scale generative media systems, image and video foundation models, production AI infrastructure, and inference optimization — bridging frontier research with systems that ship to global products.
Led the end-to-end development and deployment of Meta AI’s production lipsync generation system, enabling images to sing to music using state-of-the-art generative video models.
Directed post-training and productionization of MovieGen-based image-to-video and video-to-video systems deployed across Meta AI, Instagram, Facebook, WhatsApp, and Meta’s advertising platforms.
Led the efficiency workstream for the MovieGen family of frontier video generation foundation models, reducing generation latency from tens of minutes to seconds.
Led the technical strategy and execution of Imagine Flash, Meta’s real-time image generation system integrated into Meta AI and WhatsApp.
My research spans generative media, diffusion and flow models, inference optimization, 3D vision, and computer vision. Selected work includes Movie Gen, Imagine Flash, Bespoke Solvers, Avatars Grow Legs, Re-ReND, VisCo Grids, ASSANet, and DeepGCNs.
View research highlights →Before Meta, I led computer vision and deep learning research organizations at KAUST, working with PhD students, postdoctoral researchers, junior research scientists, and academic and industrial collaborators. I have published in leading AI conferences including CVPR, ICCV, ECCV, NeurIPS, ICML, and ICLR.
Read more →