Krishna Somandepalli
Krishna Somandepalli is a Staff Research Engineer at Google DeepMind in New York City who works on multimodal AI, from audio and facial expression to generative video. He earned a master's degree in electrical and computer engineering at UC Santa Barbara in 2012, worked on brain connectomes with functional MRI at NYU Langone Medical Center until 2015, and began a PhD at the University of Southern California in 2015 under Shri Narayanan. At USC's Signal Analysis and Interpretation Laboratory, he led the MICA group, collaborated with the Geena Davis Institute on Project SeeJane, and interned at Google from 2018 to 2020. The 2023 Richard Deswarte Prize in Digital History recognised his research on Auschwitz-Birkenau survivor narratives, and VideoPoet, a language model for zero-shot video generation that he co-authored, won an ICML 2024 Best Paper award. He contributes to the Gemini 2.5 family of models, with a focus on understanding audio and human expression.
- Now at
- Google DeepMind
- Badges
Timeline
-
2007–2012
Engineering degrees in India and California
He earns a B.E. in electronics and communication engineering at University Visvesvaraya College of Engineering in 2011. In 2012, he completes an M.S. in electrical and computer engineering at the University of California, Santa Barbara, advised by Matthew Turk.
-
2012–2015
Brain imaging at NYU Langone
From 2012 to 2015, he works as a junior research scientist at NYU Langone Medical Center. There he quantifies subcortical brain connectomes with functional MRI in children with autism.
-
2015
A PhD at USC
In 2015, he begins a PhD in electrical and computer engineering at the University of Southern California, advised by Shri Narayanan in the Signal Analysis and Interpretation Laboratory. His thesis is titled Learning Shared Subspaces across Multiple Views and Modalities.
-
By 2020
Media intelligence at SAIL
By 2020, he leads the MICA group at SAIL, helps create the Center for Computational Media Intelligence, and builds three large-scale open-source datasets. With the Geena Davis Institute, he works on Project SeeJane, which analyzes over 600 movies from 2014 to 2019.
-
2018–2020
Three Google research internships
He interns at Google in 2018, 2019 and 2020. In 2020, he demonstrates federated learning for classifying 42 emotion labels at scale in a privacy-preserving manner.
-
2023
Deswarte Prize in Digital History
He receives the 2023 Richard Deswarte Prize in Digital History for research developing computational methods to analyze Auschwitz-Birkenau survivor narratives.
-
2024
VideoPoet wins ICML Best Paper
He is a co-author of VideoPoet, a decoder-only transformer language model that learns across video, image, audio and text and synthesizes video with matching audio. The paper receives a Best Paper award at ICML 2024.
-
As of October 2026
Staff Research Engineer at DeepMind
He is a Staff Research Engineer at Google DeepMind in New York City, after earlier work at Google Research. He contributes to the Gemini 2.5 family, with a focus on multimodal capabilities, particularly understanding audio and human expression.
Career
- NowGoogle DeepMindStaff Research Engineer
- 2018–2020GoogleIntern
- 2012–2015NYU Langone Medical CenterJunior Research Scientist
- University of Southern CaliforniaMICA Group Leader
Education
- From 2015University of Southern CaliforniaPhD
- 2012University of California, Santa BarbaraM.S.
- 2011University Visvesvaraya College of EngineeringB.E.
Sources and credits
Photo: We have no published photo of Krishna Somandepalli yet. Send us a free-licensed photo.
