Ananda Theertha Suresh
Also known as Ananda Suresh
Ananda Theertha Suresh is a machine-learning researcher whose work connects information theory, federated learning and language-model inference. He shared a 2015 NIPS Best Paper award for work on Good-Turing distribution estimation and visited the Simons Institute's Information Theory program as a graduate student. By 2017, he held a BTech from IIT Madras and a PhD in electrical and computer engineering from UC San Diego under Alon Orlitsky. He coauthored research on distributed mean estimation and SCAFFOLD for federated learning, along with a 2025 method that speeds speculative decoding by verifying token blocks together. As of October 2026, he is a research scientist at Google Research, developing algorithms for machine-learning systems with resource and privacy considerations.
- Now at
- Google Research
- Links
- Website and Google Scholar
Timeline
-
2015
A NIPS best paper
He shares the 2015 NIPS Best Paper award with his UC San Diego adviser Alon Orlitsky for “Competitive Distribution Estimation: Why is Good-Turing Good.”
-
2015
Visiting the Simons Institute
During spring 2015, he visits the Simons Institute's Information Theory program as a Visiting Graduate Student.
-
By 2017
Degrees at IIT and UCSD
By 2017, he holds a BTech in Engineering Physics from IIT Madras and a PhD in electrical and computer engineering from UC San Diego under Alon Orlitsky.
-
2017
Two research honors
In 2017, he receives an ICML honorable mention for “A Unified Maximum Likelihood Approach for Estimating Symmetric Properties of Discrete Distributions,” and the Marconi Society names him a Paul Baran Young Scholar.
-
2017
Distributed mean estimation
In 2017, he coauthors “Distributed Mean Estimation with Limited Communication”; a structured random rotation before quantization reduces mean-estimation error for distributed clients.
-
2020
SCAFFOLD for federated learning
In 2020, he coauthors “SCAFFOLD: Stochastic Controlled Averaging for Federated Learning,” which uses control variates to correct client drift.
-
2025
Block level verification
In 2025, he coauthors “Block level verification accelerates speculative decoding,” which verifies drafted token blocks jointly to speed large language model inference.
-
As of October 2026
Research scientist at Google
As of October 2026, he is a research scientist at Google Research. He develops algorithms and theoretical formulations for machine-learning systems with resource and privacy considerations. His research spans machine learning, information theory, and statistics.
Career
- NowGoogle ResearchResearch Scientist
- 2015Simons Institute for the Theory of ComputingVisiting Graduate Student
Education
Sources and credits
Photo: We have no published photo of Ananda Theertha Suresh yet. Send us a free-licensed photo.
- NIPS 2015 Awards
- ECE Student, Professor Win Best Paper Award at NIPS Conference
- Ananda Suresh
- Marconi Society Honors Ananda Theertha Suresh with The Paul Baran Young Scholar Award - The Marconi Society
- Ananda Theertha Suresh
- Ananda Theertha Suresh, 2017
- ICML 2017 Awards
- Distributed Mean Estimation with Limited Communication
- SCAFFOLD: Stochastic Controlled Averaging for Federated Learning
- Block level verification accelerates speculative decoding
- Ananda Theertha Suresh

