PeopleofSuperintelligence

Researcher

Eli Tran-Johnson

Eli Tran-Johnson is an AI researcher and systems engineer whose work spans language-model self-evaluation, AI feedback, model testing and coding tools. In 2022, he coauthored research on model self-evaluation and Constitutional AI and managed Anthropic's research cluster behind Softmax Linear Units. In 2023, he coauthored research on model-written evaluations and moral self-correction and held the title of Senior Systems Engineer at Anthropic. As of October 2026, he is an Anthropic colleague whose work includes naming clide, an internal command-line tool for code editing with Claude.

Timeline

  1. 2022

    Models judge their answers

    In 2022, he coauthors Language Models (Mostly) Know What They Know, which tests whether language models can evaluate their own claims and predict which questions they will answer correctly.

  2. 2022

    Constitutional AI

    In 2022, he coauthors Constitutional AI: Harmlessness from AI Feedback, a method that trains a harmless AI assistant through self-improvement guided by a list of rules or principles.

  3. 2022

    Managing the research cluster

    In 2022, he and Nova DasSarma manage Anthropic's research cluster, which supports the Softmax Linear Units study.

  4. 2023

    Model-written evaluations

    In 2023, he coauthors a study that uses model-written evaluations to generate 154 datasets and finds cases where language models get worse as they grow in size.

  5. 2023

    Moral self-correction

    In 2023, he coauthors research testing whether language models trained with reinforcement learning from human feedback can be instructed to avoid harmful outputs; the paper finds moral self-correction emerges at 22B parameters.

  6. September 2023

    Senior systems engineer

    In September 2023, he is a Senior Systems Engineer at Anthropic and says his team uses future reservations to plan capacity for its first open beta product launch.

  7. As of October 2026

    Anthropic colleague

    As of October 2026, he is an Anthropic colleague, and his work includes naming clide, an internal command-line tool for chatting with Claude to edit code and complete development tasks.

  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 Eli Tran-Johnson yet. Send us a free-licensed photo.

  1. Language Models (Mostly) Know What They Know arxiv.org. Checked 19 October 2026.
  2. Constitutional AI: Harmlessness from AI Feedback arxiv.org. Checked 19 October 2026.
  3. Softmax Linear Units transformer-circuits.pub. Checked 19 October 2026.
  4. Discovering Language Model Behaviors with Model-Written Evaluations aclanthology.org. Checked 19 October 2026.
  5. The Capacity for Moral Self-Correction in Large Language Models arxiv.org. Checked 19 October 2026.
  6. Using Compute Engine future reservations for capacity planning cloud.google.com. Checked 19 October 2026.
  7. The Making of Claude Code canwinai.cn. Checked 19 October 2026.