PeopleofSuperintelligence

Researcher

Suriya Gunasekar

Suriya Gunasekar is a machine learning researcher and one of the authors behind Microsoft's Phi family of small language models. They received a PhD from the University of Texas at Austin in 2016, studied the implicit bias of gradient descent in papers from 2017 and 2018, and were a research assistant professor at the Toyota Technological Institute at Chicago before joining Microsoft Research. In 2023, Gunasekar headed the author list of Textbooks Are All You Need, which introduced the 1.3B-parameter code model phi-1, and went on to coauthor the Phi-3 and Phi-4 technical reports in 2024 and the Phi-4-reasoning report in 2025. They are a Principal Research Manager at Microsoft Research and part of the Physics of AGI group that pioneered the Phi models.

Works
43
Citations
1,457
h-index
18
Badges

Publication numbers from OpenAlex, as of 22 October 2026.

Timeline

  1. 2016

    PhD at UT Austin

    They receive a PhD from the University of Texas at Austin in 2016, advised by Joydeep Ghosh.

  2. 2017–2018

    Implicit bias of gradient descent

    In 2017, they coauthor Implicit Regularization in Matrix Factorization with Blake Woodworth, Srinadh Bhojanapalli, Behnam Neyshabur and Nathan Srebro, which conjectures that gradient descent on a full-dimensional factorization converges to the minimum nuclear norm solution. A 2018 paper with Jason Lee, Daniel Soudry and Nathan Srebro shows that gradient descent on full-width linear convolutional networks converges to a predictor related to a bridge penalty in the frequency domain.

  3. 2019

    Visiting scientist at Simons Institute

    In 2019, they join the Foundations of Deep Learning program at the Simons Institute for the Theory of Computing as a visiting scientist. They hold a research assistant professor post at the Toyota Technological Institute at Chicago before moving to Microsoft Research.

  4. 2023

    Textbooks Are All You Need

    In 2023, they head the author list of Textbooks Are All You Need, which introduces phi-1, a Transformer-based language model for code with 1.3B parameters, trained for 4 days on 8 A100s.

  5. December 2023

    The Phi-2 small language model

    Microsoft Research publishes its Phi-2 announcement in December 2023, with them among the named contributors. On complex benchmarks, Phi-2 matches or outperforms models up to 25x larger.

  6. 2024

    Phi-3 and Phi-4 reports

    They coauthor the Phi-3 Technical Report in 2024, which introduces phi-3-mini, a 3.8 billion parameter language model trained on 3.3 trillion tokens. The Phi-4 Technical Report of December 2024, also with them as an author, presents phi-4, a 14-billion parameter model built with a training recipe centered on data quality.

  7. April 2025

    The Phi-4-reasoning model

    The Phi-4-reasoning Technical Report appears in April 2025 with them among its authors. It introduces Phi-4-reasoning, a 14-billion parameter reasoning model trained by supervised fine-tuning of Phi-4.

  8. As of October 2026

    Principal Research Manager at Microsoft

    They are a Principal Research Manager at Microsoft Research and part of the Physics of AGI group that pioneered the Phi family of language models. Their work focuses on improving language model capabilities through scale and data curation.

Career

  1. NowMicrosoft ResearchPrincipal Research Manager
  2. 2019Simons Institute for the Theory of ComputingVisiting Scientist
  3. 2019–2024Microsoft
  4. 2018–2020Toyota Technological Institute at ChicagoResearch Assistant Professor

Education

  1. 2016University of Texas at AustinPhD

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

  1. Among employers with 30+ full-time cases, OpenAI leads primary technical role H-1B pay at $310,000, and Amazon, Google and Microsoft file 57% of all casesIncludes Microsoft. Across 440,669 certified US H-1B filings from 44 employers (FY2020–FY2026), OpenAI leads primary technical role base pay at $310,000 and xAI leads overall base pay at $290,000.
  2. Google had the most people shaping AI every year from 2014 to 2025Includes Microsoft. 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.
  3. The company share of AI researchers’ top papers was 31% in 2020–2024 and 8% in 1970–1974Includes Microsoft. We estimate where the people shaping AI worked when their most-cited papers appeared: the company share was 31% in 2020–2024 and 8% in 1970–1974, and Google led all organizations in 2020–2024.
  4. Google has the highest combined h-index among AI organizationsIncludes Microsoft. 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 Suriya Gunasekar yet. Send us a free-licensed photo.

  1. Suriya Gunasekar simons.berkeley.edu. Checked 20 October 2026.
  2. Implicit Regularization in Matrix Factorization arxiv.org. Checked 20 October 2026.
  3. Implicit Bias of Gradient Descent on Linear Convolutional Networks arxiv.org. Checked 20 October 2026.
  4. Suriya Gunasekar sgunasekar.github.io. Checked 20 October 2026.
  5. Textbooks Are All You Need arxiv.org. Checked 20 October 2026.
  6. Phi-2: The surprising power of small language models microsoft.com. Checked 20 October 2026.
  7. Phi-3 Technical Report: A Highly Capable Language Model Locally on Your Phone arxiv.org. Checked 20 October 2026.
  8. Phi-4 Technical Report microsoft.com. Checked 20 October 2026.
  9. Phi-4-reasoning Technical Report microsoft.com. Checked 20 October 2026.