PeopleofSuperintelligence

Researcher

Mayee Chen

Also known as Mayee F. Chen

Mayee Chen is a machine learning researcher who works on improving models through their training data, and she is a founding scientist at Engram. After graduating summa cum laude from Princeton, she began a computer science PhD at Stanford in 2019 in Christopher Ré's Hazy Research Lab, where her early work on weak supervision included the FlyingSquid framework and a UAI 2022 Best Student Paper Runner-Up Award. Her later papers focused on language models: the Skill-it framework in 2023, the Aioli data-mixing method in 2024 and the Weaver verification system in 2025. As a research intern at the Allen Institute for AI from 2025 to 2026, she drove data mixing for the Olmo open models, was a core contributor to Olmo 3 and published the Olmix framework in 2026.

Now at
Engram
Works
30
Citations
274
h-index
8
Badges

Publication numbers from OpenAlex, as of 22 October 2026.

Timeline

  1. 2015–2019

    Operations research at Princeton

    She studies at Princeton from 2015 to 2019 and graduates summa cum laude with a concentration in Operations Research and Financial Engineering (ORFE) and a certificate in Applications of Computing. From January 2018 to May 2019, she does research with Miklos Racz in Princeton ORFE, and she also works with Elad Hazan. She interns at Google in 2017 and works as a quantitative trading intern at IMC in 2018.

  2. 2019–2020

    FlyingSquid and weak supervision

    She begins a computer science PhD at Stanford in 2019, advised by Christopher Ré in the Hazy Research Lab, and focuses on improving machine learning through data. With Daniel Y. Fu and other coauthors, she builds FlyingSquid, a weak supervision framework described in a February 2020 paper that recovers model parameters 170 times faster on average than previous approaches.

  3. 2022

    Paper awards at UAI and AAAI

    Her paper "Shoring Up the Foundations: Fusing Model Embeddings and Weak Supervision," written with Daniel Y. Fu and others, wins a Best Student Paper Runner-Up Award at UAI 2022. In 2022, "The Details Matter," her work on preventing class collapse in supervised contrastive learning, wins a Best Paper Award at the AAAI Workshop on Artificial Intelligence with Biased or Scarce Data.

  4. 2023

    The Skill-it framework

    In July 2023, she and her coauthors introduce Skill-it!, a framework built on the idea that language models learn skills from their training data in a natural order, along with an online data sampling algorithm over mixtures of skills. The paper is a spotlight at NeurIPS 2023, and she is a research intern at Microsoft in 2023.

  5. 2024–2025

    Aioli and Smoothie

    In November 2024, she releases Aioli with Michael Y. Hu, Nicholas Lourie, Kyunghyun Cho and Christopher Ré; the method treats existing data-mixing methods as solving one optimization problem and estimates mixing-law parameters during training to adjust data proportions. Aioli appears at ICLR 2025, and she coauthors Smoothie, a label-free method for routing among language models, at NeurIPS 2024.

  6. 2025

    Weaver and weak verifiers

    In June 2025, she, Jon Saad-Falcon and E. Kelly Buchanan present Weaver, which combines weak verifiers such as reward models and LM judges without labeled data and distills them into a compact 400M model. Its paper, "Shrinking the Generation-Verification Gap with Weak Verifiers," appears at NeurIPS 2025.

  7. 2025–2026

    Data mixing for Olmo 3

    From 2025 to 2026, she is a research intern at the Allen Institute for AI, driving data mixing for its Olmo open-source language models, and she is a core contributor to Olmo 3, a family of fully open 7B and 32B models described in December 2025. In February 2026, she publishes Olmix with coauthors including Luca Soldaini and Kyle Lo; its mixture-reuse method matches full recomputation of the mix with 74% less compute. Olmix is an oral presentation at the Data-FM workshop at ICLR 2026, and she presents it at Snorkel AI's San Francisco office in 2026.

  8. Since 2026

    Founding scientist at Engram

    She joins Engram as a founding scientist in 2026. In August 2026, Harvey credits her as an Engram contributor to Calderwood & Harkness, a synthetic law firm built with Engram to test how well AI agents search, reason over and use institutional knowledge.

Career

  1. Since 2026EngramFounding scientist
  2. 2025–2026Allen Institute for AI (Ai2)Research intern
  3. 2023MicrosoftResearch intern
  4. 2018–2019Princeton UniversityResearcher
  5. 2018IMCQuantitative trading intern
  6. 2017GoogleIntern

Education

  1. 2019Princeton UniversityStudent
  2. From 2019Stanford UniversityPhD

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

Most-cited works

  1. Ask Me Anything: A simple strategy for prompting language models2022 · 84 citationsWith Christopher Ré
  2. Fast and Three-rious: Speeding Up Weak Supervision with Triplet Methods2020 · 39 citationsWith Christopher Ré
  3. An Adversarial Model of Network Disruption: Maximizing Disagreement and Polarization in Social Networks2021 · IEEE Transactions on Network Science and Engineering · 22 citations
  4. Mandoline: Model Evaluation under Distribution Shift2021 · 21 citationsWith Christopher Ré
  5. DataComp-LM: In search of the next generation of training sets for language models2024 · neural information processing systems · 19 citationsWith Kyle Lo, Luke Zettlemoyer, Matt Jordan, Niklas Muennighoff, Pang Wei Koh and Shuran Song

From OpenAlex, as of 21 October 2026.

  1. The Massachusetts Institute of Technology trained or employed 44 of the 187 AI major-prize winners, more than any other universityIncludes Mayee Chen. The Massachusetts Institute of Technology trained or employed 44 of the 187 AI major-prize winners, more than any other university. We compare doctorate and 5+ year employment links, career timing and current workplaces for 187 major-prize and 1,111 best-paper winners.
  2. 16% of AI career moves in the 2020s went from a university to a companyIncludes Mayee Chen. We estimate that 16% of AI career moves in the 2020s went from a university to a company; 29% of moves starting at a university went to a company, and 76% ended the university role.
  3. 20% of people at companies who shape AI also hold a university role, led by Google and Stanford UniversityIncludes Stanford University. We estimate that 539 of the 6,766 people with a known current organization (8%) hold both a university role and a company role, including 20% of people at companies and 13% of people at universities.
  4. Google had the most people shaping AI every year from 2014 to 2025Includes Google. 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.

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 Mayee Chen yet. Send us a free-licensed photo.

  1. Hello, I’m Mayee! - Mayee Chen mayeechen.github.io. Checked 20 October 2026.
  2. Mayee F Chen (Engram): machine learning robustness — Lacuna lacuna.tiptreesystems.com. Checked 20 October 2026.
  3. Miklos Z. Racz racz.statistics.northwestern.edu. Checked 20 October 2026.
  4. [2002.11955] Fast and Three-rious: Speeding Up Weak Supervision with Triplet Methods arxiv.org. Checked 20 October 2026.
  5. [2307.14430] Skill-it! A Data-Driven Skills Framework for Understanding and Training Language Models arxiv.org. Checked 20 October 2026.
  6. [2411.05735] Aioli: A Unified Optimization Framework for Language Model Data Mixing arxiv.org. Checked 20 October 2026.
  7. Weaver: Closing the Generation-Verification Gap with Weak Verifiers · Hazy Research hazyresearch.stanford.edu. Checked 20 October 2026.
  8. [2512.13961] Olmo 3 arxiv.org. Checked 20 October 2026.
  9. [2602.12237] Olmix: A Framework for Data Mixing Throughout LM Development arxiv.org. Checked 20 October 2026.
  10. Understanding Olmix: A Framework for Data Mixing Throughout Language Model Development | Snorkel AI snorkel.ai. Checked 20 October 2026.
  11. LAB: Law Firm Knowledge | Harvey harvey.ai. Checked 20 October 2026.