PeopleofSuperintelligence

Builder

Akshita Bhagia

Akshita Bhagia is a senior research engineer at AI2 who builds and evaluates large language models. She earned a master's degree in computer science at the University of Massachusetts Amherst in 2020 and coauthored studies of multimodal text generation and instruction tuning. Her work includes OLMo and the Dolma training corpus, followed by FlexOlmo and Olmo 3. In April 2026, she coauthored Ai2 research on modular post-training with mixture-of-experts models.

Works
22
Citations
185
h-index
7

Publication numbers from OpenAlex, as of 18 October 2026.

Timeline

  1. 2020

    A master's at UMass

    She earns a master's degree in computer science from the University of Massachusetts Amherst in 2020, where she works with Mohit Iyyer on natural language processing and digital humanities.

  2. 2022–2023

    Text generation and HINT

    In 2022, she coauthors a study of image-conditioned text generation and self-rationalization across three multimodal tasks. In 2023, she coauthors HINT, which turns task instructions and examples into parameter-efficient modules for language models.

  3. 2024

    OLMo and Dolma

    In 2024, she coauthors OLMo, an open language model released with open training data, training code, and evaluation code. In 2024, she also coauthors Dolma, a three-trillion-token English corpus for language-model pretraining.

  4. August 2025

    Building FlexOlmo

    In August 2025, she coauthors FlexOlmo, which lets data owners train on private datasets and decide when their data is active in the model.

  5. December 2025

    The Olmo 3 model family

    In December 2025, she coauthors Olmo 3, a fully open family of language models at 7B and 32B parameter scales.

  6. As of October 2026

    Modular model post-training

    In April 2026, she coauthors Ai2's report on modular post-training with mixture-of-experts models. As of October 2026, she is a Senior Research Engineer on the AllenNLP team at Ai2 and builds and evaluates large language models.

A language model matched the links in this timeline to people, organizations and papers (provisional, model-judged).

Career

  1. Since 2022Allen Institute for AI (Ai2)Senior Research Engineer

Education

  1. 2020University of Massachusetts AmherstMaster's degree in computer science

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

Most-cited works

  1. OLMo: Accelerating the Science of Language Models2024 · Annual Meeting of the Association for Computational Linguistics (ACL) · 69 citationsWith Luke Zettlemoyer, Nathan Lambert, Niklas Muennighoff and Noah Smith
  2. Dolma: an Open Corpus of Three Trillion Tokens for Language Model Pretraining Research2024 · Annual Meeting of the Association for Computational Linguistics (ACL) · 56 citationsWith Luke Zettlemoyer, Nathan Lambert, Niklas Muennighoff and Noah Smith
  3. What's In My Big Data?2023 · 9 citationsWith Noah Smith
  4. OLMoE: Open Mixture-of-Experts Language Models2024 · 8 citationsWith Ali Farhadi, Nathan Lambert, Niklas Muennighoff, Noah Smith and Pang Wei Koh
  5. Continued Pretraining for Better Zero- and Few-Shot Promptability2022 · Conference on Empirical Methods in Natural Language Processing (EMNLP) · 7 citations

From OpenAlex, as of 18 October 2026.

  1. The company share of AI researchers’ top papers rose from 2% to 33%Includes Allen Institute for AI (Ai2). We estimate where leading AI researchers worked when their most-cited papers appeared. The company share rose from 2% in 1975–1979 to 33% in 2020–2024.
  2. Google has the highest combined h-index among AI organizationsIncludes Allen Institute for AI (Ai2). Our estimates of combined and median h-index at AI organizations, using current affiliations and OpenAlex records.
  3. 44% of AI career moves go from one university to anotherIncludes Allen Institute for AI (Ai2). We estimate how 11,906 AI career moves link universities, companies, labs and government research centers, with net flows and the most common paths.
  4. Stanford University trained or employed 27 of the 113 AI major-prize winners, more than any other universityIncludes University of Massachusetts Amherst. Stanford University trained or employed 27 of the 113 AI major-prize winners, more than any other university. We compare doctorate and employment links for major-prize and best-paper winners.

Colleagues

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

Sources and credits

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

  1. Akshita Bhagia akshitab.github.io. Checked 18 October 2026.
  2. On Advances in Text Generation from Images Beyond Captioning: A Case Study in Self-Rationalization - ACL Anthology aclanthology.org. Checked 18 October 2026.
  3. HINT: Hypernetwork Instruction Tuning for Efficient Zero- and Few-Shot Generalisation - ACL Anthology aclanthology.org. Checked 18 October 2026.
  4. OLMo: Accelerating the Science of Language Models - ACL Anthology aclanthology.org. Checked 18 October 2026.
  5. Dolma: an Open Corpus of Three Trillion Tokens for Language Model Pretraining Research - ACL Anthology aclanthology.org. Checked 18 October 2026.
  6. Ai2 Newsletter | August 2025 allenai.org. Checked 18 October 2026.
  7. Olmo 3 arxiv.org. Checked 18 October 2026.
  8. Train separately, merge together: Modular post-training with mixture-of-experts | Ai2 allenai.org. Checked 18 October 2026.