PeopleofSuperintelligence

Lab leader

Sam McCandlish

Also known as Samuel McCandlish

Sam McCandlish is a co-founder of Anthropic, a company formed to promote AI safety, and he works there as chief architect. He completed a PhD at Stanford University in 2017 with a dissertation titled Depth Perception in Holography. He submitted Scaling Laws for Neural Language Models to arXiv in January 2020, and Language Models are Few-Shot Learners received an Outstanding Paper award at NeurIPS in 2020. Constitutional AI: Harmlessness from AI Feedback, a paper he co-authored, was submitted to arXiv in December 2022. In October 2025 he moved from the CTO role to chief architect.

Timeline

  1. 2017

    A holography dissertation

    At Stanford University he completes a PhD in 2017 with the dissertation Depth Perception in Holography.

  2. January 2020

    Scaling laws for language models

    In January 2020 he submits Scaling Laws for Neural Language Models to arXiv.

  3. 2020

    An outstanding paper award

    Language Models are Few-Shot Learners, a paper he co-authors, wins an Outstanding Paper award at NeurIPS in 2020.

  4. January 2021

    Co-founding Anthropic

    He co-founds Anthropic, which former members of OpenAI start in January 2021 to promote AI safety.

  5. 2022

    Constitutional AI and RLHF

    He is an author of Training a Helpful and Harmless Assistant with Reinforcement Learning from Human Feedback, submitted in April 2022. Constitutional AI: Harmlessness from AI Feedback, submitted in December 2022, also lists him as an author.

  6. October 2025

    Moving to chief architect

    In October 2025 he takes on the chief architect role, after serving as Anthropic's CTO.

  7. As of October 2026

    Chief architect at Anthropic

    He is co-founder and chief architect of Anthropic, and he focuses on large-scale model training.

Sources and credits

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

  1. Samuel McCandlish - The Mathematics Genealogy Project genealogy.math.ndsu.nodak.edu. Checked 14 October 2026.
  2. [2001.08361] Scaling Laws for Neural Language Models arxiv.org. Checked 14 October 2026.
  3. NeurIPS 2020 Awards neurips.cc. Checked 14 October 2026.
  4. [2005.14165] Language Models are Few-Shot Learners arxiv.org. Checked 14 October 2026.
  5. Leadership at Anthropic \ Anthropic anthropic.com. Checked 14 October 2026.
  6. Anthropic - Wikipedia en.wikipedia.org. Checked 14 October 2026.
  7. [2204.05862] Training a Helpful and Harmless Assistant with Reinforcement Learning from Human Feedback arxiv.org. Checked 14 October 2026.
  8. [2212.08073] Constitutional AI: Harmlessness from AI Feedback arxiv.org. Checked 14 October 2026.
  9. Rahul Patil joins as Chief Technology Officer \ Anthropic anthropic.com. Checked 14 October 2026.