PeopleofSuperintelligence

Researcher

Dmitry Lepikhin

Also known as Dima Lepikhin

Dmitry Lepikhin is a machine-learning researcher whose work helped scale large language models. He worked on Google's Sibyl project and co-founded Scaled Inference with Olcan Sercinoglu in 2014. His research included GShard and GLaM for model scaling, dialog-focused LaMDA and multimodal Gemini. He co-authored Gemma 3 in 2025 and Gemma 4's technical report in 2026; as of October 2026, he is a Google researcher focused on language-model scaling.

Now at
Google
Works
15
Citations
3,200
h-index
11
Badges

Publication numbers from OpenAlex, as of 19 October 2026.

Timeline

  1. By 2014

    Sibyl at Google

    By 2014, his Google work includes Sibyl, a project used in products such as YouTube and Google Play.

  2. 2014

    Co-founding Scaled Inference

    With Olcan Sercinoglu, he founds Scaled Inference in 2014.

  3. 2020

    GShard scales translation

    In 2020, he co-authors GShard, which scales multilingual machine translation beyond 600 billion parameters through conditional computation and automatic sharding.

  4. 2021

    The GLaM model

    In 2021, he co-authors GLaM, a sparsely activated mixture-of-experts language model with 1.2 trillion parameters.

  5. 2022

    LaMDA for dialog

    In 2022, he co-authors LaMDA, a family of Transformer language models specialized for dialog.

  6. 2023

    Gemini's multimodal model family

    In 2023, he co-authors Gemini's technical report on a family of highly capable multimodal models.

  7. 2025

    Gemma 3 open models

    In 2025, he co-authors Gemma 3, a multimodal family of lightweight open models ranging from 1 to 27 billion parameters.

  8. As of October 2026

    Google researcher

    He is a researcher at Google focused on scaling language models, mixture-of-experts architectures and machine-learning systems. In 2026, he co-authors the Gemma 4 technical report, which introduces open-weight, natively multimodal language models.

Career

  1. Since 2021GoogleResearcher
  2. From 2014Scaled InferenceCo-founder

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

Most-cited works

  1. Gemini: A Family of Highly Capable Multimodal Models2023 · 988 citationsWith Angeliki Lazaridou, Anja Hauth, Clemens Meyer, David Silver, Emily Pitler and Fabio Viola
  2. LaMDA: Language Models for Dialog Applications2022 · 778 citationsWith Apoorv Kulshreshtha, Heng-Tze Cheng, Meredith Ringel Morris, Quoc Le and Ray Kurzweil
  3. GShard: Scaling Giant Models with Conditional Computation and Automatic Sharding2020 · 411 citations
  4. Massively Multilingual Neural Machine Translation in the Wild: Findings and Challenges2019 · 281 citationsWith Colin Cherry, Melvin Johnson and Yonghui Wu
  5. GLaM: Efficient Scaling of Language Models with Mixture-of-Experts2021 · 237 citationsWith Quoc Le

From OpenAlex, as of 19 October 2026.

  1. Google has the highest combined h-index among AI organizationsIncludes Google. Our estimates of combined and median h-index at AI organizations, using current affiliations and OpenAlex records.
  2. Google alumni founded or led 87 other organizations, the most among the labs we compareIncludes Google. We trace 212 organizations founded or led by AI lab alumni, and 223 career moves between the labs.
  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 from 1961 to 2025, with a year-by-year chart.
  4. Of 462 people shaping AI who worked at Google, 190 work there nowIncludes Google. Our estimates of where former Google and Google DeepMind people work now, including moves within Google.

Colleagues

People with dated positions at the same organization in the same years.

Sources and credits

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

  1. Backed By Tencent And Felicis, Scaled Inference Wants To Be The Google Brain For Everyone techcrunch.com. Checked 19 October 2026.
  2. GShard: Scaling Giant Models With Conditional Computation and Automatic Sharding research.google. Checked 19 October 2026.
  3. GLaM: Efficient Scaling of Language Models with Mixture-of-Experts arxiv.org. Checked 19 October 2026.
  4. LaMDA: Language Models for Dialog Applications research.google. Checked 19 October 2026.
  5. Gemini: A Family of Highly Capable Multimodal Models arxiv.org. Checked 19 October 2026.
  6. Gemma 3 Technical Report arxiv.org. Checked 19 October 2026.
  7. Dmitry Lepikhin | alphaXiv alphaxiv.org. Checked 19 October 2026.
  8. Gemma 4 Technical Report arxiv.org. Checked 19 October 2026.