PeopleofSuperintelligence

Researcher

David Heineman

David Heineman is a language-model researcher whose work spans text-generation evaluation, decoding, and open models. At Georgia Tech, he coauthored LENS and SALSA for text simplification and helped build Thresh for fine-grained text evaluation. His later work included multi-prompt decoding, studies of benchmark reliability, and Ai2's OLMo 3 report. He also worked on language-model pretraining with Meta Superintelligence Labs (FAIR). As of October 2026, he is a first-year PhD student at Stanford, focused on pretraining, data, and language-model evaluation.

Works
12
Citations
8
h-index
2

Publication numbers from OpenAlex, as of 19 October 2026.

Timeline

  1. 2023

    LENS for text simplification

    At Georgia Tech, he works with Wei Xu, Yao Dou, and Mounica Maddela; in 2023, he coauthors LENS, a learnable metric for text simplification.

  2. 2023

    SALSA for text simplification

    In 2023, he coauthors SALSA, an edit-based human-annotation framework for evaluating text simplification.

  3. 2023

    Thresh evaluation platform

    In 2023, he coauthors Thresh, a unified and customizable platform for fine-grained evaluation of generated text.

  4. 2024

    Multi-prompt decoding

    In 2024, he coauthors a decoding method that uses multiple prompts and Minimum Bayes Risk to choose among candidate generations.

  5. 2025

    Signal and Noise

    In 2025, he coauthors Signal and Noise, a study of benchmark signal and noise, and Fluid Benchmarking, which selects evaluation items dynamically for each language model.

  6. 2025

    OLMo 3

    At Ai2, he works on the OLMo team with advisers Kyle Lo and Jesse Dodge; in 2025, he coauthors OLMo 3, a family of fully open language models.

  7. As of October 2026

    A Stanford PhD student

    Before Stanford, he works on language-model pretraining with Meta Superintelligence Labs (FAIR). As of October 2026, he is a first-year PhD student at Stanford, working on pretraining, data and evaluation for language models.

Career

  1. 2025Allen Institute for AI (Ai2)
  2. 2023Georgia Institute of Technology

Education

  1. From 2026Stanford UniversityPhD

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

Most-cited works

  1. Olmo 32025 · 5 citationsWith Akari Asai, Allyson Ettinger, Aman Rangapur, Amanda Bertsch, Ashish Sabharwal and Bailey Kuehl
  2. Terminal-Bench: Benchmarking Agents on Hard, Realistic Tasks in Command Line Interfaces2026 · 2 citationsWith Andy Konwinski, Nicholas Carlini and Niklas Muennighoff
  3. Improving Minimum Bayes Risk Decoding with Multi-Prompt2024 · Conference on Empirical Methods in Natural Language Processing (EMNLP) · 1 citations
  4. DataDecide: How to Predict Best Pretraining Data with Small Experiments2025 · 0 citationsWith Noah Smith and Pang Wei Koh
  5. Establishing Task Scaling Laws via Compute-Efficient Model Ladders2024 · 0 citationsWith Noah Smith and Pang Wei Koh

From OpenAlex, as of 19 October 2026.

  1. Google alumni founded or led 87 other organizations, the most among the labs we compareIncludes Meta. We trace 212 organizations founded or led by AI lab alumni, and 223 career moves between the labs.
  2. Stanford University trained or employed 20 of the 89 AI major-prize winners, more than any other universityIncludes Georgia Institute of Technology. Stanford University trained or employed 20 of the 89 AI major-prize winners, more than any other university. We compare doctorate and employment links for major-prize and best-paper winners.
  3. The company share of AI researchers’ top papers rose from 2% to 32%Includes Meta. We estimate where leading AI researchers worked when their most-cited papers appeared. The company share rose from 2% in 1975–1979 to 32% in 2020–2024.
  4. 44% of AI career moves go from one university to anotherIncludes Meta. We estimate how 14,272 AI career moves link universities, companies, labs and government research centers, with net flows and the most common paths.

Colleagues

People with dated positions at the same organization in the same years.

Sources and credits

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

  1. David Heineman davidheineman.com. Checked 19 October 2026.
  2. ACL 2023 | Georgia Tech – Research Impact & Leadership sites.gatech.edu. Checked 19 October 2026.
  3. LENS: A Learnable Evaluation Metric for Text Simplification - ACL Anthology aclanthology.org. Checked 19 October 2026.
  4. Dancing Between Success and Failure: Edit-level Simplification Evaluation using SALSA - ACL Anthology aclanthology.org. Checked 19 October 2026.
  5. Thresh: A Unified, Customizable and Deployable Platform for Fine-Grained Text Evaluation - ACL Anthology aclanthology.org. Checked 19 October 2026.
  6. Improving Minimum Bayes Risk Decoding with Multi-Prompt - ACL Anthology aclanthology.org. Checked 19 October 2026.
  7. Signal and Noise: Reducing uncertainty in language model evaluation | Ai2 allenai.org. Checked 19 October 2026.
  8. Fluid Language Model Benchmarking arxiv.org. Checked 19 October 2026.
  9. Olmo 3 arxiv.org. Checked 19 October 2026.