Ishan Misra
2024 Recent Alumni Achievement Award
Ishan Misra is a research scientist in computer vision and machine learning with a focus on scalable algorithms that learn with minimal supervision. Currently, he works at Meta’s GenAI Research, leading a team at Meta working on open research problems in generative AI and multimodal learning.
Ishan’s work both at CMU and at Meta into self-supervised learning for computer vision has changed the way vision models are trained. While the standard method of supervised learning is dependent on massive labeled (by humans) datasets, Ishan has offered a scalable way to train models by visual data alone, offering the potential to greatly expand the types of problems AI can solve. His results were key to establishing self-supervised learning as a promising alternative to the dominant supervised-learning paradigm both in academia and industry. With self-supervised learning, future AI models have the potential to be generally more intelligent and less biased and to continue improving.
Ishan’s techniques form the basis of multiple different recognition systems at Meta and other companies. More recently, Ishan's work in generative video modeling is at the core of Meta's generative video systems.
At Carnegie Mellon, Ishan earned his master of science in 2014 and doctorate in 2018 in robotics with advisors Martial Hebert, now dean of the School of Computer Science, and Professor of Robotics Abhinav Gupta.
He was recognized as one of MIT Technology Review's 35 Innovators Under 35, an annual list of emerging industry leaders who are developing new technologies or using existing ones creatively to solve problems with a global impact. Ishan has authored over 50 research papers at top-tier venues for AI research and is regularly invited to present his research to prestigious universities and industry research labs.