PeopleofSuperintelligence

Researcher

Stephen Roller

Stephen Roller is a researcher on large language models and dialogue systems. A computer science PhD from the University of Texas at Austin under Katrin Erk, he worked at Meta's Fundamental AI Research lab (FAIR) from 2017 to 2023. At FAIR he was a joint first author of Wizard of Wikipedia in 2019, first author of a 2020 paper on recipes for open-domain chatbots, and a joint first author of OPT, a suite of open transformers of up to 175 billion parameters, in 2022. After FAIR, he led pretraining and training infrastructure efforts at Character.AI from 2023 to 2024, then worked as a Senior Staff Research Scientist at Google DeepMind from 2024 to 2025 and is a listed author of the 2025 Gemini 2.5 report. At Thinking Machines Lab, where he was named among 29 employees announced in February 2025, he is a Member of the Technical Staff.

Works
50
Citations
3,982
h-index
25
Badges

Publication numbers from OpenAlex, as of 22 October 2026.

Timeline

  1. 2010–2017

    PhD at UT Austin

    He earns a computer science PhD at the University of Texas at Austin under Katrin Erk, working from 2010 to 2017 on distributional semantics, multimodal models and Bayesian methods. His 2017 dissertation is Identifying Lexical Relationships and Entailments with Distributional Semantics.

  2. 2017–2023

    Research engineer at FAIR

    He joins Meta's Fundamental AI Research lab (FAIR) in 2017 and stays until 2023. On the Large Language Model and dialogue teams, he works as a Senior Staff Research Engineer, and much of his FAIR work is open source.

  3. 2019

    Wizard of Wikipedia

    In 2019, he is a joint first author, with Emily Dinan and Kurt Shuster, of Wizard of Wikipedia, published at ICLR. The paper releases a large dataset of conversations grounded in Wikipedia knowledge and builds dialogue models that look up that knowledge, read it and answer with it.

  4. 2020

    Recipes for an open-domain chatbot

    In April 2020, he is first author of Recipes for building an open-domain chatbot, which builds models with 90M, 2.7B and 9.4B parameters and makes the models and code publicly available. Human evaluations rate the best models above existing approaches for engagingness and humanness.

  5. 2022

    The OPT language models

    In 2022, he is a joint first author, with Susan Zhang and Naman Goyal, of OPT: Open Pre-trained Transformer Language Models. OPT is a suite of decoder-only transformers from 125M to 175B parameters; OPT-175B is comparable to GPT-3 with 1/7th the carbon footprint, and the team releases a logbook of its infrastructure challenges along with code.

  6. 2023–2024

    Pretraining at Character.AI

    He moves to Character.AI as a Member of the Technical Staff in 2023 and leads its pretraining and training infrastructure efforts until 2024.

  7. 2024–2025

    Google DeepMind research scientist

    From 2024 to 2025, he is a Senior Staff Research Scientist at Google DeepMind. In July 2025, the Gemini 2.5 report appears on arXiv with him among its authors.

  8. Since 2025

    Joining Thinking Machines Lab

    In February 2025, Thinking Machines Lab announces a team of 29 employees that includes him. He is a Member of the Technical Staff at the lab.

Career

  1. Since 2025Thinking Machines LabMember of the Technical Staff
  2. 2024–2025Google DeepMindSenior Staff Research Scientist
  3. 2023–2024Character.AIMember of the Technical Staff
  4. 2023–2024New York University
  5. 2018–2024Meta
  6. 2017–2023FAIR (Meta)Senior Staff Research Engineer

Education

  1. 2017University of Texas at AustinPhD

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

  1. The median h-index is 48 at universities and 14 at frontier AI labsIncludes Stephen Roller. 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. Google alumni have founded or led 133 other organizations, more than any other AI labIncludes Stephen Roller. We estimate that alumni of 17 major AI labs went on to found or lead 350 other organizations, including 226 they founded, and we trace 472 career moves directly between 23 labs since 2010.
  3. Of 943 people shaping AI who worked at Google, 446 work there nowIncludes Stephen Roller. Our estimates of where the 943 people shaping AI who worked at Google and Google DeepMind work now: 446 at Google and 452 across 278 other organizations.
  4. 23% of people shaping AI who left a university for a company later returned to a universityIncludes Stephen Roller. We estimate that 468 of the 2,011 people shaping AI who left a university for a company (23%) later returned to a university after a median of 5 years, and 163 of 829 (20%) made the reverse round trip.

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 Stephen Roller yet. Send us a free-licensed photo.

  1. Stephen Roller stephenroller.com. Checked 20 October 2026.
  2. AI-Lab People - Stephen Roller - UT Austin Computer Science cs.utexas.edu. Checked 20 October 2026.
  3. Stephen Roller on alphaXiv alphaxiv.org. Checked 20 October 2026.
  4. Wizard of Wikipedia: Knowledge-Powered Conversational agents arxiv.org. Checked 20 October 2026.
  5. Recipes for building an open-domain chatbot arxiv.org. Checked 20 October 2026.
  6. OPT: Open Pre-trained Transformer Language Models arxiv.org. Checked 20 October 2026.
  7. Gemini 2.5: Pushing the Frontier with Advanced Reasoning, Multimodality, Long Context, and Next Generation Agentic Capabilities arxiv.org. Checked 20 October 2026.
  8. Former OpenAI CTO Mira Murati unveils Thinking Machines Lab details and leadership team stacked with former OpenAI colleagues fortune.com. Checked 20 October 2026.