PeopleofSuperintelligence

Researcher

Aitor Ormazabal

Aitor Ormazabal is a researcher at Reka whose work spans natural language processing and multimodal AI. He studied cross-lingual word embeddings and coauthored work on language models for poetry, paraphrase generation, and black-box model adaptation. He completed a doctorate on controllable NLP in 2024 and coauthored Reka's report on the Core, Flash, and Edge models. His 2025 and 2026 papers explored visually augmented language models and multimodal skill composition.

Now at
Reka AI
Works
18
Citations
112
h-index
5
Badges

Publication numbers from OpenAlex, as of 18 October 2026.

Timeline

  1. 2015–2020

    Studying mathematics and AI

    From 2015 to 2019, he studies at the University of the Basque Country and earns a mathematics bachelor's degree at Universidad Nacional de Educación a Distancia. He completes a master's in artificial intelligence at the University of Edinburgh in 2020. In 2019, he and coauthors examine limitations in cross-lingual word-embedding mappings.

  2. 2021

    Context anchoring for embeddings

    In 2021, he and coauthors introduce context anchoring, which learns source-language embeddings against fixed target-language embeddings using translated context words as anchors.

  3. 2022

    Poetry and paraphrase systems

    In 2022, he coauthors PoeLM, a language model that uses control codes for poetic meter and rhyme, and a parallel-corpus method for paraphrase generation.

  4. 2023

    Adapting black-box language models

    In 2023, he and coauthors present CombLM, which adapts large language models to new domains and tasks by combining a fine-tuned small model with a black-box model.

  5. 2024

    A doctorate in NLP control

    In 2024, he defends his doctorate at the University of the Basque Country with a thesis on general attribute controllability in NLP models.

  6. 2024

    Reka's multimodal models

    In 2024, he coauthors Reka's report on Core, Flash, and Edge, multimodal language models trained to process text, images, video, and audio.

  7. 2025–2026

    Visually augmented and multimodal models

    In 2025, he and coauthors present BLIND-VALM, a visually augmented language model that uses visually grounded text representations. In 2026, they study how multimodal language models compose skills across modalities.

  8. Since 2024

    Researcher at Reka

    He works as a researcher at Reka AI from 2024 onward.

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

Career

  1. Since 2024Reka AIResearcher

Education

  1. 2024University of the Basque CountryDoctorate
  2. 2020University of EdinburghMaster's in Artificial Intelligence
  3. 2019University of the Basque Country
  4. 2019National University of Distance EducationMathematics bachelor's degree

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

Most-cited works

  1. Analyzing the Limitations of Cross-lingual Word Embedding Mappings2019 · Annual Meeting of the Association for Computational Linguistics (ACL) · 61 citationsWith Eneko Agirre
  2. PoeLM: A Meter- and Rhyme-Controllable Language Model for Unsupervised Poetry Generation2022 · 19 citationsWith Eneko Agirre
  3. Beyond Offline Mapping: Learning Cross Lingual Word Embeddings through Context Anchoring2020 · 8 citationsWith Eneko Agirre
  4. Latxa: An Open Language Model and Evaluation Suite for Basque2024 · Annual Meeting of the Association for Computational Linguistics (ACL) · 8 citationsWith Eneko Agirre
  5. Principled Paraphrase Generation with Parallel Corpora2022 · Annual Meeting of the Association for Computational Linguistics (ACL) · 8 citationsWith Eneko Agirre

From OpenAlex, as of 18 October 2026.

  1. The company share of AI researchers’ top papers rose from 2% to 32%Includes Reka AI. We estimate where leading AI researchers worked when their most-cited papers appeared. The company share rose from 2% in 1975–1979 to 32% in 2020–2024.
  2. Google alumni founded or led 79 other organizations, the most among the labs we compareIncludes Reka AI. We trace 181 organizations founded or led by AI lab alumni, and 187 career moves between the labs.
  3. Stanford University trained or employed 27 of the 112 AI major-prize winners, more than any other universityIncludes University of Edinburgh. Stanford University trained or employed 27 of the 112 AI major-prize winners, more than any other university. We compare doctorate and employment links for major-prize and best-paper winners.
  4. Google has the highest combined h-index among AI organizationsIncludes University of Edinburgh. Our estimates of combined and median h-index at AI organizations, using current affiliations and OpenAlex records.

Colleagues

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

Sources and credits

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

  1. Aitor Ormazabal — Lacuna lacuna.tiptreesystems.com. Checked 18 October 2026.
  2. Analyzing the Limitations of Cross-lingual Word Embedding Mappings aclanthology.org. Checked 18 October 2026.
  3. Beyond Offline Mapping: Learning Cross-lingual Word Embeddings through Context Anchoring aclanthology.org. Checked 18 October 2026.
  4. PoeLM: A Meter- and Rhyme-Controllable Language Model for Unsupervised Poetry Generation aclanthology.org. Checked 18 October 2026.
  5. Principled Paraphrase Generation with Parallel Corpora arxiv.org. Checked 18 October 2026.
  6. CombLM: Adapting Black-Box Language Models through Small Fine-Tuned Models arxiv.org. Checked 18 October 2026.
  7. 2024_02_29; 10:30 Doctoral Thesis Defence Aitor Ormazabal Oregi ehu.eus. Checked 18 October 2026.
  8. Reka Core, Flash, and Edge: A Series of Powerful Multimodal Language Models arxiv.org. Checked 18 October 2026.
  9. Improving the Efficiency of Visually Augmented Language Models aclanthology.org. Checked 18 October 2026.
  10. Multimodal LLMs Do Not Compose Skills Optimally across Modalities aclanthology.org. Checked 18 October 2026.
  11. Alumni - Eneko Agirre eagirre.github.io. Checked 18 October 2026.