Ishan Misra

His father retired as head of the genetics department at a Pune research institute, his mother is a chemistry M.Sc. and teacher. In 2008 he began a B.Tech. in computer science at IIIT Hyderabad and later won the gold medal for the highest CGPA.⁠[1]

In September 2013, as a master’s student at Carnegie Mellon UniversityCarnegie Mellon UniversityCarnegie Mellon University (CMU) is a private research university in Pittsburgh, Pennsylvania. It grew out of the Carnegie Technical Schools, founded with funding from Andrew Carnegie in 1900, which were renamed the Carnegie Institute of Technology in 1912 and took the present name after merging with the Mellon Institute in 1967. The university is known for computer science, robotics, engineering, and the arts, and has campuses in Qatar, Silicon Valley, and Rwanda. As of September 2025, it enrolled 16,582 students. Its current president is Farnam Jahanian. On September 30, 2026, the university announced a $3 billion gift from Ken Griffin and plans to build a new campus in Miami.Open the full entry’s Robotics Institute, he was named to the 2014 class of Siebel Scholars, each receiving a $35,000 award; his research focused on object detection and scene understanding, using data-driven methods to learn object detectors without heuristics or human intervention.⁠[2]

He earned a master’s and a PhD in robotics from Carnegie Mellon University in 2014 and 2018, advised by Martial Hebert and Abhinav Gupta.⁠[3]

His dissertation, “Visual Learning with Minimal Human Supervision,” was runner-up for CMU’s 2018 SCS Distinguished Dissertation Award; that year he joined Facebook New York as a research scientist, continuing work on self-supervised learning.⁠[1]

With Laurens van der Maaten he published Pretext-Invariant Representation Learning (PIRL) at CVPR 2020, arguing that semantic representations should be invariant under image transformations; the paper said it set new state-of-the-art results on several self-supervised benchmarks and better object-detection representations than supervised pre-training.⁠[4]

In March 2021 he and co-authors including Priya Goyal and Mathilde Caron published SEER, a 1.3-billion-parameter RegNetY model trained without supervision on 1 billion random, uncurated images that reached 84.2% top-1 accuracy on ImageNet.⁠[5]

On July 31, 2021, as a FAIR research scientist, he was the guest of episode 206 of the Lex FridmanLex FridmanLex Fridman is a Soviet-born, US-raised computer scientist and podcast host who works as an MIT research scientist on human-AI interaction and autonomous driving. His Lex Fridman Podcast is known for hours-long conversations with guests including heads of state and technology leaders.Open the full entry Podcast, titled “Self-Supervised Deep Learning in Computer Vision,” covering self-supervised learning, contrastive learning, data augmentation, SwAV, SEER and the VISSL library.⁠[6]

In 2022, as a research scientist at Meta AI Research, he was named to MIT Technology Review’s 35 Innovators Under 35 list for research showing models can be trained on visual data alone.⁠[7]

He co-authored the CVPR 2023 paper ImageBind, which learns a joint embedding across six modalities—images, text, audio, depth, thermal and IMU data—and shows that image-paired data alone suffices to bind them together.⁠[8]

In November 2023 Meta debuted Emu Video, which generates four-second animated clips at 512×512 and 16 frames per second from a caption or image.⁠[9]

He co-authored the Emu Video paper; the model first generates an image from text and then a video conditioned on the text and image, and in human evaluations its videos were preferred 81% of the time over GoogleGoogleGoogle is an American technology company founded by Larry Page and Sergey Brin and incorporated in 1998. It became an Alphabet subsidiary in 2015. Its businesses include internet search, advertising, software platforms, cloud computing, and artificial intelligence.Open the full entry’s Imagen Video.⁠[10]

He received Carnegie Mellon University’s 2024 Recent Alumni Achievement Award; the university said his self-supervised learning research changed how vision models are trained and his generative video work is at the core of Meta’s generative video systems.⁠[3]

In October 2024 Meta’s team published Movie Gen, a cast of foundation models that generate 1080p HD video with synchronized audio and support video editing and personalization, the largest video model being a 30-billion-parameter transformer; he is one of the paper’s authors.⁠[11]

As of October 2026 he is a director and research scientist in the TBD Labs division of Meta’s superintelligence group; he previously led video generation research in Meta’s GenAI group as overall tech lead of Movie Gen, and received a 2025 PAMI Young Researcher Award honorable mention.⁠[12]

Personal website: Public link: https://imisra.github.io/ (opens in a new window)⁠[6]

X: Public account: https://x.com/imisra_ (opens in a new window)⁠[6]