AI People Wiki

Builders & open source

Sasha Rush

ML researcher; former Cornell and Harvard professor and former Cursor and Hugging Face researcher

Sasha Rush (Alexander M. Rush in paper bylines) is an ML researcher, programmer and blogger. Rush was previously a professor at Harvard and then Cornell, worked as an ML researcher at Cursor on the Composer series of models, and worked at Hugging Face on open-source LLMs. Rush is a co-author of the paper that introduced Hugging Face's Transformers library and of the Zephyr alignment paper, wrote The Annotated Transformer (2018), and helped start the Conference on Language Modeling in 2024.1234

At a glance

Current role
None confirmed in the sources
Sector
Company
Education
Cornell, Harvard
Work listed
3
Honors
5
Sources cited
7
Profile reviewed

Career

Rush completed a PhD in computer science at MIT in 2014, advised by Michael Collins, with a dissertation on relaxation methods for natural language decoding, and then held a post-doctoral fellowship at Facebook AI Research in 2015. Rush was an assistant professor of computer science at Harvard from 2015 to 2019 and then an associate professor of computer science at Cornell University.5

From 2019 Rush also worked with Hugging Face, where the team built open-source LLMs. Rush later worked as an ML researcher at Cursor, helping develop the Composer series of models, and is a listed author of the Composer 2 technical report from Cursor Research.156

Research

Rush's academic group worked on early LLMs, natural language processing and infrastructure. Papers on Rush's list include "A Neural Attention Model for Abstractive Sentence Summarization" (EMNLP 2015) with Sumit Chopra and Jason Weston, "Character-Aware Neural Language Models" (AAAI 2016), "Sequence-Level Knowledge Distillation" (EMNLP 2016) and "Compound Probabilistic Context-Free Grammars for Grammar Induction" (ACL 2019).15

More recent work covers language models without attention, including "Pretraining Without Attention" (EMNLP 2023 Findings) and MambaByte (COLM 2024), as well as "Simple and Effective Masked Diffusion Language Models" (NeurIPS 2024) and "Contextual Document Embeddings" (ICLR 2025).15

Open-source work and community

Rush co-authored the OpenNMT toolkit for neural machine translation (ACL Demo 2017) and the LSTMVis tool for inspecting recurrent network hidden states (InfoVis 2017), wrote The Annotated Transformer (2018) and published GPU-Puzzles in 2021. Rush helped start the Conference on Language Modeling (COLM), served as a Senior Program Chair and a General Chair of ICLR.157

Work and contributions

  • Transformers Published research

    Co-author of the paper introducing Hugging Face's Transformers library (arXiv 2019; EMNLP 2020 demos), listed on Rush's homepage.31

  • Zephyr Published research

    Co-author of "Zephyr: Direct Distillation of LM Alignment" (2023; COLM 2024), on aligning a smaller chat model with distilled direct preference optimization.41

  • The Annotated Transformer Published research

    Sole author of The Annotated Transformer, presented at the ACL 2018 NLP-OSS workshop, per Rush's publication list.1

Timeline

  1. Completed a PhD in computer science at MIT, advised by Michael Collins.5
  2. Held a post-doctoral fellowship at Facebook AI Research and joined Harvard as an assistant professor of computer science.5
  3. Published "A Neural Attention Model for Abstractive Sentence Summarization" at EMNLP with Sumit Chopra and Jason Weston.5
  4. Co-authored OpenNMT, an open-source toolkit for neural machine translation, presented at ACL Demo.15
  5. Wrote The Annotated Transformer, presented at the ACL NLP-OSS workshop.1
  6. Joined Cornell University as an associate professor of computer science and began working with Hugging Face.5
  7. Co-authored the paper introducing Hugging Face's Transformers library.3
  8. Co-authored "Zephyr - Direct Distillation of LM Alignment" with the Hugging Face team.4
  9. Listed as an author of the Composer 2 technical report from Cursor Research.6

Documented roles

The sources do not confirm a current role; these are the roles they document.

Honors and recognition

  • NSF CAREER Award · National Science Foundation, 20195
  • Sloan Research Fellowship · Alfred P. Sloan Foundation, 20215
  • Best Paper Award, EMNLP · Conference on Empirical Methods in Natural Language Processing, 20105
  • Best Paper Award, NAACL · North American Chapter of the Association for Computational Linguistics, 20125
  • Outstanding Paper, EMNLP · Conference on Empirical Methods in Natural Language Processing, 20235

Companies in their history

Suggest a correction or removal

Profiles cover public professional work only, and each statement cites a source. If something is out of date, missing or wrong, tell us and link a source. People can also ask for their profile to be corrected or removed.

References

7 sources · profile reviewed Oct 21, 2026 · data updated Oct 21, 2026

  1. Sasha Rush — homepage (bio and selected publications)Sasha Rush · Official page · accessed Oct 21, 2026
  2. srush (Sasha Rush) on GitHubGitHub · Official page · accessed Oct 21, 2026
  3. HuggingFace's Transformers: State-of-the-art Natural Language Processing (arXiv 1910.03771)arXiv · Paper · published Oct 9, 2019 · accessed Oct 21, 2026
  4. Zephyr: Direct Distillation of LM Alignment (arXiv 2310.16944)arXiv · Paper · published Oct 25, 2023 · accessed Oct 21, 2026
  5. Alexander M. Rush — curriculum vitae (PDF)Sasha Rush · Official page · accessed Oct 21, 2026
  6. Composer 2 Technical Report (arXiv 2603.24477)arXiv · Paper · published Mar 25, 2026 · accessed Oct 21, 2026
  7. Conference on Language Modeling (COLM)COLM · Official page · accessed Oct 21, 2026