Kannan Ramchandran

Kannan Ramchandran is an electrical engineer and computer scientist at UC Berkeley who brings coding and information theory to distributed machine learning. After a B.E. at the City College of New York in 1982, he worked at AT&T Bell Labs from 1984 to 1990, completed a Ph.D. at Columbia in 1993 and taught at the University of Illinois before moving to Berkeley's EECS department in 1999. There he coauthored a 2008 paper introducing regenerating codes for distributed storage and a 2015 paper on speeding up distributed machine learning with codes, and he received best paper awards from the IEEE Information Theory Society and Communication Society in 2012 and 2020. He later coauthored work on Byzantine-robust distributed learning in 2018, clustered federated learning in 2020 and SPEX, a 2025 method for explaining the outputs of large language models. A winner of the 2017 IEEE Kobayashi Computers and Communications Award, he holds the Gilbert Henry Gates Endowed Chair of Electrical Engineering and Computer Sciences and is affiliated with Berkeley Artificial Intelligence Research.

Now at: University of California, Berkeley
Works: 715
Citations: 32,103
h-index: 86
Badges: Bridge
Loyal

Publication numbers from OpenAlex, as of 22 October 2026.

Timeline

1982–1990

Columbia degrees and Bell Labs

He earns a B.E. from the City College of New York in 1982 and an M.S. from Columbia University in 1984. From 1984 to 1990, he works at AT&T Bell Labs.
1993–1999

Faculty at Illinois

He completes his Ph.D. at Columbia in 1993 and becomes an assistant professor in the ECE department at the University of Illinois at Urbana-Champaign and associate director of the Beckman Institute Multimedia Laboratory. He stays on the Illinois faculty until 1999 and receives a Hank Magnuski Scholar award there in 1998.
1999–2009

The BASiCS Lab at Berkeley

He joins UC Berkeley's EECS department in 1999 and creates the Berkeley Audiovisual Signal Processing and Communication Systems (BASiCS) Lab. He becomes an IEEE Fellow in 2005 for contributions to image and video communications, and he wins an Outstanding Teaching Award at Berkeley in 2009.
2008–2017

Regenerating codes for storage

In 2008, he coauthors "Network Coding for Distributed Storage Systems" with Alexandros Dimakis, Brighten Godfrey, Yunnan Wu and Martin Wainwright; it introduces regenerating codes, which reduce the bandwidth needed to repair a failed storage node. He receives a best paper award from the IEEE Information Theory Society and Communication Society in 2012. In 2017, he receives the IEEE Koji Kobayashi Computers and Communications Award for contributions to the theory and practice of distributed source and storage coding.
2015–2020

Codes for distributed machine learning

In December 2015, he coauthors "Speeding Up Distributed Machine Learning Using Codes" with Kangwook Lee, Maximilian Lam, Ramtin Pedarsani and Dimitris Papailiopoulos, using codes to offset slow straggler workers in distributed matrix multiplication and to cut communication in data shuffling. The IEEE Information Theory Society names him Padovani Lecturer in 2019, and he receives the 2020 Communications Society and Information Theory Society Joint Paper Award.
2018–2020

Robust and federated learning

At ICML 2018, he, Dong Yin, Yudong Chen and Peter Bartlett present distributed learning algorithms, built on median and trimmed-mean gradient descent, that are provably robust when some computing units behave adversarially. With Avishek Ghosh, Jichan Chung and Dong Yin, he proposes the Iterative Federated Clustering Algorithm (IFCA) for clustered federated learning, with preliminary results at NeurIPS 2020.
2025

SPEX for explaining LLMs

He coauthors SPEX (Spectral Explainer), published at ICML 2025 with Justin Singh Kang, Bin Yu and others. The model-agnostic method applies a sparse Fourier transform with a channel decoding algorithm to find the input interactions that shape a large language model's output on long-context tasks.
As of October 2026

Gates Chair at Berkeley

He is the Gilbert Henry Gates Endowed Chair of Electrical Engineering and Computer Sciences at UC Berkeley, with research affiliations that include Berkeley Artificial Intelligence Research (BAIR) and the CLIMB center for the theoretical foundations of learning. His interests span adversarial and distributed machine learning, coded computing, and privacy and security, and he teaches Probability and Random Processes (EECS 126) in Spring 2026.

Career

Since 1999University of California, BerkeleyGilbert Henry Gates Endowed Chair of Electrical Engineering and Computer Sciences, Founder
1993–1999University of Illinois Urbana-ChampaignAssistant professor
1984–1990Bell Labs