PeopleofSuperintelligence

Researcher

Julia Kreutzer

Julia Kreutzer is a computational linguistics researcher whose work connects machine translation, multilingual language models and evaluation. She coauthored research on bandit learning for neural machine translation in 2017 and eBay user feedback in 2018. By 2020, she had completed a PhD in Computational Linguistics at Heidelberg and worked on Google Translate in Montreal; her thesis included Joey NMT. At Google Research, she worked on machine translation with Masakhane on low-resource African languages in 2021, then coauthored the Aya 23 report in 2024 and Déjà Vu on multilingual LLM evaluation in 2025. In June 2026, she coauthored a survey about cultural gaps in AI; as of October 2026, she is a Staff Research Scientist at Cohere Labs and coauthored work on in-language reasoning across 60 languages.

Now at
Cohere

Timeline

  1. 2017

    Bandit feedback for translation

    In 2017, she coauthors a study that lifts bandit learning to neural sequence-to-sequence models with attention-based recurrent networks. On a neural machine translation task, simulated feedback improves domain adaptation by up to 5.89 BLEU points.

  2. 2018

    User feedback for translation

    In 2018, she coauthors a neural machine translation study using explicit and implicit feedback collected on eBay. The team finds that implicit feedback from cross-lingual search improves task-specific and translation-quality metrics.

  3. 2020

    A PhD and Joey NMT

    In 2020, she completes a PhD in Computational Linguistics at Heidelberg University's Statistical NLP Group and works as a research scientist in Google Translate's Montreal team. Her thesis work includes Joey NMT, a clean and minimalistic code base that helps machine-translation novices.

  4. 2021

    Google Research and Masakhane

    In 2021, she researches machine translation as a Google Research scientist in Montreal, using deep and reinforcement learning and human-in-the-loop methods. She also builds translation models for low-resource African languages with the Masakhane community.

  5. 2024

    Aya 23 in 23 languages

    In 2024, she coauthors the Aya 23 technical report. The report introduces a family of multilingual language models serving 23 languages.

  6. April 2025

    Multilingual LLM evaluation

    In April 2025, she coauthors Déjà Vu, which applies machine-translation evaluation practices to multilingual LLMs. The study identifies evaluation gaps and offers a checklist of recommendations for multilingual LLM research.

  7. June 2026

    Cultural gaps in global AI

    In June 2026, she coauthors a survey on cultural gaps in AI as a Senior Research Scientist at Cohere. The survey finds that many users switch from their primary language to English and report inadequate answers due to gaps in cultural understanding.

  8. As of October 2026

    Staff Research Scientist at Cohere Labs

    As of October 2026, she is a Staff Research Scientist at Cohere Labs, working on multilingual and multicultural LLMs. In October 2026, she coauthors research on data mixing for in-language reasoning; Tiny Aya L2-Thinker reasons in the prompt language over 93% of the time across 60 languages.

Career

  1. NowCohereSenior Research Scientist
  2. From 2021Google ResearchResearch Scientist
  3. Until 2020GoogleResearch Scientist

Education

  1. Heidelberg UniversityPhD

From the biography.

  1. Google alumni have founded or led 133 other organizations, more than any other AI labIncludes Julia Kreutzer. We estimate that alumni of 17 major AI labs went on to found or lead 350 other organizations, including 226 they founded, and we trace 472 career moves directly between 23 labs since 2010.
  2. Of 943 people shaping AI who worked at Google, 446 work there nowIncludes Julia Kreutzer. Our estimates of where the 943 people shaping AI who worked at Google and Google DeepMind work now: 446 at Google and 452 across 278 other organizations.
  3. Among employers with 30+ full-time cases, OpenAI leads primary technical role H-1B pay at $310,000, and Amazon, Google and Microsoft file 57% of all casesIncludes Cohere. Across 440,669 certified US H-1B filings from 44 employers (FY2020–FY2026), OpenAI leads primary technical role base pay at $310,000 and xAI leads overall base pay at $290,000.
  4. Google had the most people shaping AI every year from 2014 to 2025Includes Google. Our estimates of which organizations employed the most people shaping AI each year from 1955 to 2025, tracing the shift from university labs to companies.

Colleagues

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

Sources and credits

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

  1. Bandit Structured Prediction for Neural Sequence-to-Sequence Learning - ACL Anthology aclanthology.org. Checked 21 October 2026.
  2. Can Neural Machine Translation be Improved with User Feedback? - ACL Anthology aclanthology.org. Checked 21 October 2026.
  3. Reinforcement learning with human feedback for neural machine translation europe.naverlabs.com. Checked 21 October 2026.
  4. InterNLP2021 - Invited Speakers sites.google.com. Checked 21 October 2026.
  5. Aya 23: Open Weight Releases to Further Multilingual Progress arxiv.org. Checked 21 October 2026.
  6. Déjà Vu: Multilingual LLM Evaluation through the Lens of Machine Translation Evaluation arxiv.org. Checked 21 October 2026.
  7. Why Cultural Awareness is Essential for Global AI cohere.com. Checked 21 October 2026.
  8. About Me | Julia Kreutzer juliakreutzer.github.io. Checked 21 October 2026.
  9. Multilingual Bridges: How Data Mixing Unlocks In-Language Reasoning cohere.com. Checked 21 October 2026.