Builders & open source
Hannaneh Hajishirzi
Torode Family Professor, Paul G. Allen School of Computer Science & Engineering; leads the H2Lab, University of Washington
- Torode Family Professor, Paul G. Allen School of Computer Science & Engineering; leads the H2Lab, University of Washington12
- Senior Director; co-lead of OLMo and Tülu, Ai2 (Allen Institute for AI)12
Hannaneh (Hanna) Hajishirzi is the Torode Family Professor in the Paul G. Allen School of Computer Science & Engineering at the University of Washington, where Hajishirzi leads the H2Lab. Hajishirzi is also a Senior Director at Ai2, co-leading the OLMo and Tülu projects. Hajishirzi's research covers large language models, reasoning and agents, open models and data, the science of AI and evaluation, and AI for science. Hajishirzi is the last-listed author of the OLMo and Tülu 3 papers and a co-author of Dolma, and the OLMoE paper credits Hajishirzi with direction and advising of the overall effort.123456
At a glance
- Current roles
- Torode Family Professor, Paul G. Allen School of Computer Science & Engineering; leads the H2Lab +1 more
- Companies
- University of Washington, Ai2
- Sector
- University
- Fields
- Language models
- Education
- Carnegie Mellon, UIUC, UW
- Work listed
- 4
- Honors
- 5
- Sources cited
- 11
- Profile reviewed
Career
The Allen School's faculty page records that Hajishirzi earned a PhD in computer science at the University of Illinois at Urbana-Champaign and then spent a year as a postdoctoral associate at Disney Research and Carnegie Mellon University. At the University of Washington, Hajishirzi holds the Torode Family professorship in Computer Science & Engineering, with adjunct appointments in Electrical & Computer Engineering and Linguistics, and leads the H2Lab.2
Hajishirzi's homepage presents a concurrent role as Senior Director at Ai2, where Hajishirzi co-leads the OLMo and Tülu projects on fully open language and reasoning models. The team spans research scientists, engineers and young investigators at Ai2 alongside students and postdocs at H2Lab.1
Research
Hajishirzi's research covers large language models, reasoning and agents, open AI initiatives, the science of AI and evaluation, and AI for science. Earlier work in machine reading includes Bidirectional Attention Flow (BiDAF), a question-answering model published at ICLR 2017 with Minjoon Seo, Aniruddha Kembhavi and Ali Farhadi.17
With students and collaborators, Hajishirzi co-authored Self-Instruct (ACL 2023), which aligns language models using instructions they generate themselves, and Self-RAG (2023), which trains a model to decide when to retrieve passages and to critique its own output.89
Open models and data
At Ai2, Hajishirzi is a co-author of the Dolma pretraining corpus paper (accepted at ACL 2024) and the last-listed author of the 2024 OLMo paper, which released an open language model with its training data and code. The OLMoE paper (2024) credits Hajishirzi with direction and advising of the overall effort, and Hajishirzi is also an author of "2 OLMo 2 Furious" (2024) and the Tülu 3 post-training report (2024).436105
Work and contributions
OLMo Open weights
Co-led OLMo, per Hajishirzi's homepage; last-listed author of the OLMo paper (2024), and credited in OLMoE with direction and advising of the overall effort.136
Tülu Published research
Co-led Tülu, per Hajishirzi's homepage, and last-listed author of "Tulu 3: Pushing Frontiers in Open Language Model Post-Training" (2024).15
Dolma Open data
Co-author of the Dolma paper, which describes a three-trillion-token English corpus released for research on language-model pretraining.4
Olmo 3.1 32B Think Open weights
Credited in the Olmo 3 paper's author contributions for technical leadership and for mentorship and advising; the Olmo 3.1 32B Think model card cites that paper.11
Timeline
- Co-authored Bidirectional Attention Flow for Machine Comprehension, later published at ICLR 2017.7
- Received the Torode Family Career Development Professorship at the University of Washington.1
- Co-authored Self-Instruct, on aligning language models with self-generated instructions.8
- Co-authored Self-RAG, on retrieval-augmented generation with self-reflection.9
- Co-authored the Dolma open pretraining corpus paper from Ai2.4
- Last-listed author of "OLMo: Accelerating the Science of Language Models".3
- Directed and advised the OLMoE open mixture-of-experts model effort.6
- Last-listed author of the Tülu 3 open post-training report.5
- Co-authored "2 OLMo 2 Furious", the OLMo 2 report.10
Honors and recognition
Where they work
- University of Washington 19 people in the wiki
- Ai2 23 people in the wiki
Related people
- Kristina ToutanovaAlso at University of Washington
- Emily M. BenderAlso at University of Washington
- Noah A. SmithAlso at University of Washington
- Ali FarhadiAlso at University of Washington
- Yejin ChoiAlso at University of Washington
- Illia PolosukhinLanguage models
- Shixiang Shane GuLanguage models
- Stella BidermanLanguage models
- Koray KavukcuogluLanguage models
- Baptiste RozièreLanguage models
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
11 sources · profile reviewed Oct 21, 2026 · data updated Oct 22, 2026
- Hanna Hajishirzi — homepage
- Hannaneh Hajishirzi — Paul G. Allen School of Computer Science & Engineering faculty profile
- OLMo: Accelerating the Science of Language Models (arXiv 2402.00838)
- Dolma: an Open Corpus of Three Trillion Tokens for Language Model Pretraining Research (arXiv 2402.00159)
- Tulu 3: Pushing Frontiers in Open Language Model Post-Training (arXiv 2411.15124)
- OLMoE: Open Mixture-of-Experts Language Models (arXiv 2409.02060, full text with Author Contributions)
- Bidirectional Attention Flow for Machine Comprehension (arXiv 1611.01603)
- Self-Instruct: Aligning Language Models with Self-Generated Instructions (arXiv 2212.10560)
- Self-RAG: Learning to Retrieve, Generate, and Critique through Self-Reflection (arXiv 2310.11511)
- 2 OLMo 2 Furious (arXiv 2501.00656)
- Olmo 3 (arXiv 2512.13961)