PeopleofSuperintelligence

Researcher

Ahmet Üstün

Ahmet Üstün is a Cohere research scientist whose AI work spans multilingual language models, model efficiency, and coding agents. He published a Turkish morphology thesis in 2017 and earned a computational linguistics PhD from the University of Groningen in 2023. His work includes UDapter, the Aya instruction model, and studies of quantization, agentic coding, and multilingual speech. As of October 2026, he leads Cohere's Code Agents team as a research scientist and technical lead.

Now at
Cohere
Works
129
Citations
993
h-index
20
Badges

Publication numbers from OpenAlex, as of 18 October 2026.

Timeline

  1. 2017

    A Turkish morphology thesis

    In 2017, he publishes an M.S. thesis on probabilistic learning of Turkish morphosemantics.

  2. 2018

    A best-paper award

    At RepL4NLP in 2018, his paper “Characters or Morphemes: How to Represent Words?” wins a best-paper award.

  3. 2020

    UDapter, a multilingual parser

    In 2020, he, Arianna Bisazza, Gosse Bouma, and Gertjan van Noord publish UDapter, a multilingual dependency parser that learns adapters through language embeddings while sharing parameters across languages.

  4. 2021

    A NAVER research internship

    At NAVER LABS Europe, he works on adapters for multilingual unsupervised translation during a 2021 internship. He publishes “Multilingual Unsupervised Neural Machine Translation with Denoising Adapters” with Alexandre Berard, Laurent Besacier, and Matthias Gallé at EMNLP 2021.

  5. 2023

    A PhD and quantization research

    His degrees include an M.S. in Cognitive Science and a B.S. in Electrical and Electronics Engineering from Middle East Technical University. In 2023, he completes a PhD in computational linguistics at the University of Groningen with a thesis on multilingual learning and adaptation for neural language models. He coauthors a study that finds sensitivity to quantization depends on pretraining conditions.

  6. 2024

    The Aya model for 101 languages

    At Cohere For AI, he is a research scientist and a member of the Aya Open Science Initiative. In 2024, he coauthors Aya, an instruction-tuned multilingual model that follows instructions in 101 languages; the team releases its model and instruction datasets as open resources.

  7. 2026

    AI agents ignore discovered solutions

    In 2026, he coauthors a study finding that AI agents often ignore complete task solutions they discover in their environments. His work also includes Tiny Aya, a 3.35B multilingual model trained on 70 languages, and MULTISPEECHQA, a dataset with 9,200 hours and 10.8 million spoken question-answer pairs across 23 languages.

  8. As of October 2026

    Leading Cohere's Code Agents team

    In October 2026, he is a research scientist and technical lead at Cohere, where he leads the Code Agents team focused on autonomous coding systems and AI agents.

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

Career

  1. Since 2024CohereResearch Scientist
  2. 2021NAVER LABS EuropeIntern
  3. 2017–2024Bartin University
  4. 2014–2018Amasya Üniversitesi

Education

  1. 2023University of GroningenPhD
  2. Middle East Technical UniversityB.S.

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

Most-cited works

  1. UDapter: Language Adaptation for Truly Universal Dependency Parsing2020 · Conference on Empirical Methods in Natural Language Processing (EMNLP) · 87 citations
  2. An effective way of designing blended learning: A three phase design-based research approach2019 · Education and Information Technologies · 64 citations
  3. Educational UTAUT-based virtual reality acceptance scale: a validity and reliability study2022 · Virtual Reality · 62 citations
  4. The Effects of Mobile AR-based Biology Learning Experience on Students’ Motivation, Self‐Efficacy, and Attitudes in Online Learning2023 · Journal of Science Education and Technology · 60 citations
  5. Investigating the role of accepting learning management system on students’ engagement and sense of community in blended learning2021 · Education and Information Technologies · 58 citations

From OpenAlex, as of 18 October 2026.

  1. Google alumni founded or led 77 other organizations, the most among the labs we compareIncludes Cohere. We trace 177 organizations founded or led by AI lab alumni, and 186 career moves between the labs.
  2. More people shaping AI studied at Massachusetts Institute of Technology than at any other institutionIncludes University of Groningen. Our estimates of where the people shaping AI studied and where their careers led, from education records in cited biographies.
  3. The company share of AI researchers’ top papers rose from 2% to 33%Includes University of Groningen. 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.
  4. Google had the most people shaping AI every year from 2014 to 2025Our estimates of which organizations employed the most people shaping AI from 1962 to 2025, with a year-by-year chart.

Colleagues

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

Sources and credits

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

  1. About me | Ahmet Üstün ahmetustun.github.io. Checked 18 October 2026.
  2. UDapter: Language Adaptation for Truly Universal Dependency Parsing - ACL Anthology aclanthology.org. Checked 18 October 2026.
  3. Multilingual Learning and Adaptation for Neural Language Models - the University of Groningen research portal research.rug.nl. Checked 18 October 2026.
  4. Intriguing Properties of Quantization at Scale arxiv.org. Checked 18 October 2026.
  5. Aya Model: An Instruction Finetuned Open-Access Multilingual Language Model arxiv.org. Checked 18 October 2026.
  6. Aya: exploring the frontiers of Multilingual LLMs - NAVER LABS Europe europe.naverlabs.com. Checked 18 October 2026.
  7. Agents Explore but Agents Ignore: LLMs Lack Environmental Curiosity arxiv.org. Checked 18 October 2026.
  8. Tiny Aya: Bridging Scale and Multilingual Depth arxiv.org. Checked 18 October 2026.
  9. Turning Speech Language Models into Multilingual Listeners arxiv.org. Checked 18 October 2026.
  10. March Online Meetup: Ahmet Üstün from Cohere munich-nlp.com. Checked 18 October 2026.