PeopleofSuperintelligence

Researcher

Sanjiv Kumar

Sanjiv Kumar is an AI researcher whose work spans image classification, optimization, large-scale search and efficient language models. He completed a Carnegie Mellon PhD in 2005, coauthored an ICLR Best Paper on Adam in 2018, and helped develop open-source ScaNN for vector search in 2020. His later work covers efficient deep learning, language-model cascades and speculative decoding; the 2025 Faster Cascades paper received an ICLR Outstanding Paper Honorable Mention. As of October 2026, he is a Google Fellow and vice president at Google DeepMind, where he leads Gemini foundation-model work and research in retrieval, ranking and similarity search.

Works
256
Citations
13,566
h-index
49
Badges

Publication numbers from OpenAlex, as of 22 October 2026.

Timeline

  1. August 2005

    A PhD at Carnegie Mellon

    At Carnegie Mellon's Robotics Institute, he completes a PhD in August 2005 under advisor Martial Hebert. His thesis studies spatial interactions in natural images for context-based classification.

  2. 2018

    Adam convergence and an ICLR award

    With Sashank Reddi and Satyen Kale, he publishes “On the Convergence of Adam and Beyond” at ICLR in 2018. The paper gives an explicit case where Adam fails to converge on a simple convex problem, then proposes variants with long-term memory of past gradients; it wins ICLR's Best Paper Award.

  3. 2020

    ScaNN for vector search

    In July 2020, Google Research announces ScaNN, an open-source library for vector similarity search developed by a team that includes him. The team's anisotropic vector quantization improves accuracy for approximate nearest-neighbor search.

  4. February 2023

    Algorithms for efficient deep learning

    A Google Research review of efficient deep learning appears under his byline on February 7, 2023. It covers architectures, data selection, training and inference, including context-augmented language models, more efficient transformers and adaptive computation.

  5. 2024

    Language model cascades

    At ICLR in 2024, he coauthors “Language Model Cascades: Token-Level Uncertainty and Beyond.” The paper studies deferral rules for generative language models and finds that token-level uncertainty can improve their cost-quality trade-off.

  6. 2025

    Faster Cascades honorable mention

    In 2025, he coauthors “Faster Cascades via Speculative Decoding,” which combines speculative decoding with cascades for better cost-quality trade-offs and often yields higher speed-ups. ICLR gives the paper an Outstanding Paper Honorable Mention in 2025.

  7. As of October 2026

    Leading Gemini research

    As of October 2026, he is a Google Fellow and vice president at Google DeepMind, leading a machine-learning team on large foundation models, LLMs and generative AI for Gemini. He also leads research in deep retrieval and ranking, and massive-scale similarity search.

Career

  1. Since 2025Google DeepMindVice President
  2. Since 2007GoogleGoogle Fellow
  3. 2012–2026KLE Academy of Higher Education and Research
  4. 2012–2026K.L.E. Academy of Higher Education and Research Institute of Physiotherapy, Belagavi

Education

  1. 2005Carnegie Mellon UniversityPhD

From the biography and OpenAlex affiliations (the years of papers that list the organization), so some years are approximate.

Most-cited works

  1. On the Convergence of Adam and Beyond2019 · 1,771 citationsWith Satyen Kale
  2. Adaptive Federated Optimization2020 · 840 citations
  3. Semi-Supervised Hashing for Large-Scale Search2012 · IEEE Transactions on Pattern Analysis and Machine Intelligence · 811 citationsWith Shih-Fu Chang
  4. Semi-supervised hashing for scalable image retrieval2010 · IEEE Computer Society Conference on Computer Vision and Pattern Recognition · 613 citationsWith Shih-Fu Chang
  5. Learning to Hash for Indexing Big Data—A Survey2015 · Proceedings of the IEEE · 521 citationsWith Shih-Fu Chang

From OpenAlex, as of 21 October 2026.

  1. The median h-index is 48 at universities and 14 at frontier AI labsIncludes Sanjiv Kumar. The median h-index is 48 at universities and 14 at frontier AI labs. We compare publishing careers, recent citations, dual appointments and job moves across 2,420 university researchers and 308 frontier-lab researchers.
  2. The Massachusetts Institute of Technology trained or employed 44 of the 187 AI major-prize winners, more than any other universityIncludes Sanjiv Kumar. The Massachusetts Institute of Technology trained or employed 44 of the 187 AI major-prize winners, more than any other university. We compare doctorate and 5+ year employment links, career timing and current workplaces for 187 major-prize and 1,111 best-paper winners.
  3. Google had the most people shaping AI every year from 2014 to 2025Includes Google. Our estimates of which organizations employed the most people shaping AI each year from 1955 to 2025, tracing the shift from university labs to companies.
  4. Google has the highest combined h-index among AI organizationsIncludes Google. We estimate that Google holds the highest combined h-index at 8,016, while universities account for 73% of the deduplicated total across AI organizations.

Colleagues

People named in this biography, and people with dated positions at the same organization in the same years.

Sources and credits

Photo: We have no published photo of Sanjiv Kumar yet. Send us a free-licensed photo.

  1. Sanjiv Kumar - Robotics Institute Carnegie Mellon University ri.cmu.edu. Checked 22 October 2026.
  2. Sanjiv Kumar - Google DeepMind sanjivk.com. Checked 22 October 2026.
  3. On the convergence of Adam and Beyond research.google. Checked 22 October 2026.
  4. Announcing ScaNN: Efficient Vector Similarity Search research.google. Checked 22 October 2026.
  5. Google Research, 2022 & beyond: Algorithms for efficient deep learning research.google. Checked 22 October 2026.
  6. Language Model Cascades: Token-Level Uncertainty And Beyond research.google. Checked 22 October 2026.
  7. Speculative cascades — A hybrid approach for smarter, faster LLM inference research.google. Checked 22 October 2026.
  8. Announcing the Outstanding Paper Awards at ICLR 2025 blog.iclr.cc. Checked 22 October 2026.