PeopleofSuperintelligence

Researcher

Ganesh Jawahar

Ganesh Jawahar is a natural language processing researcher whose work examines how models represent language and how to make them more efficient. He coauthored a 2019 study of linguistic structure in BERT, surveyed machine-generated text detection in 2020, and coauthored AutoMoE and Orca. In April 2024, he completed a computer science PhD at the University of British Columbia and published research using GPT-4 to guide neural-architecture search. As of October 2026, he is a Research Scientist at Google DeepMind and contributes to efficiency work for on-device Gemini Nano inference.

Timeline

  1. 2019

    BERT's linguistic layers

    In 2019, Jawahar, Benoît Sagot and Djamé Seddah examine BERT's representations and find that its intermediate layers encode a hierarchy of surface, syntactic and semantic information.

  2. 2020

    Machine-generated text survey

    In 2020, Jawahar, Muhammad Abdul-Mageed and Laks V.S. Lakshmanan publish a critical survey of methods for detecting machine-generated text and analyze challenges for future research.

  3. 2022

    AutoMoE for efficient translation

    In 2022, he coauthors AutoMoE, a neural-architecture-search framework for efficient sparse mixture-of-experts Transformers in machine translation; the paper reports fourfold CPU inference speedup over manually designed Transformers.

  4. June 2023

    Orca learns from GPT-4

    In June 2023, he coauthors Orca, a 13-billion-parameter model that learns to imitate large language models' reasoning from GPT-4 explanation traces, step-by-step thought processes and complex instructions.

  5. April 2024

    A computer science PhD

    In April 2024, he completes a computer science PhD at the University of British Columbia after an MS at IIIT Hyderabad, with joint supervision from Muhammad Abdul-Mageed and Laks V.S. Lakshmanan.

  6. 2024

    GPT-4 guides architecture search

    In 2024, he coauthors a method for estimating neural-network architecture performance with GPT-4; its Hybrid-Search algorithm cuts search hours by about 50% while performing similarly to leading architecture-search methods.

  7. As of October 2026

    Google DeepMind and Gemini Nano

    He is a Research Scientist at Google DeepMind. In 2026, Gemini team developments he helped make support a method that speeds up on-device inference for Gemini Nano.

  1. Google had the most people shaping AI every year from 2014 to 2025Our estimates of which organizations employed the most people shaping AI from 1961 to 2025, with a year-by-year chart.
  2. The United States employs an estimated 53% of people in our data with a known employment countryOur estimates of where AI careers are concentrated now and how countries’ shares of dated positions changed over time.
  3. 46% of living people in our data with known birth and employment countries work outside their birth countryOur estimate of cross-border careers in AI, from country of birth and the country of each current organization.
  4. More people shaping AI studied at Massachusetts Institute of Technology than at any other institutionOur estimates of where the people shaping AI studied and where their careers led, from education records in cited biographies.

Sources and credits

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

  1. What Does BERT Learn about the Structure of Language? aclanthology.org. Checked 19 October 2026.
  2. Automatic Detection of Machine Generated Text: A Critical Survey arxiv.org. Checked 19 October 2026.
  3. AutoMoE: Heterogeneous Mixture-of-Experts with Adaptive Computation for Efficient Neural Machine Translation arxiv.org. Checked 19 October 2026.
  4. Orca - Microsoft Research: Publications microsoft.com. Checked 19 October 2026.
  5. Orca: Progressive Learning from Complex Explanation Traces of GPT-4 arxiv.org. Checked 19 October 2026.
  6. Laks V.S. Lakshmanan cs.ubc.ca. Checked 19 October 2026.
  7. LLM Performance Predictors are good initializers for Architecture Search aclanthology.org. Checked 19 October 2026.
  8. Accelerating Gemini Nano models on Pixel with frozen Multi-Token Prediction research.google. Checked 19 October 2026.