PeopleofSuperintelligence

Researcher

Kanta Yamaoka

Kanta Yamaoka is a machine learning researcher at the German Research Centre for Artificial Intelligence (DFKI) who studies bandits, preference learning and causality for sequential decision-making in personalized medicine. After working as a research assistant at DFKI from March 2022 to March 2023 and earning a BEng in computer science at Osaka Prefecture University in 2023, they spent two and a half years at Google Japan managing network and compute hardware deployments. Beginning MSc studies in computer science in winter 2025, they co-authored the Img2Vocab vocabulary-learning study in 2025, a Bayesian multi-task bandits preprint in December 2025, and the Linear-LLM-SCM causal benchmarking framework presented at an ICML 2026 workshop in July 2026. They are an MSc computer science student and research assistant in DFKI's Data Science and its Applications group, with a second affiliation at RPTU within the AI4Nof1 project.

Now at
DFKI
Badges

Timeline

  1. 2022

    Vocabulary and guided-breath learning studies

    In 2022, Yamaoka co-authors Experience is the Best Teacher: Personalized Vocabulary Building Within the Context of Instagram Posts and Sentences from GPT-3 in the UbiComp/ISWC 2022 Adjunct proceedings. A September 2022 study, Towards Guided-Breath as an Intervention Within Computer-Assisted Learning, integrates a visual breathing guide into computer-assisted learning to improve learner performance.

  2. March 2022–March 2023

    Research assistant stay at DFKI

    From March 2022 to March 2023, they live in Germany and hold a research assistant position at the German Research Centre for Artificial Intelligence (DFKI), supported by a JASSO study-abroad scholarship.

  3. 2023

    BEng at Osaka Prefecture University

    They earn a BEng in computer science at Osaka Prefecture University in Japan in 2023. Professor Koichi Kise reviews their BEng thesis, ImageToVocab: Finding New Words to Learn From One's Personal Pictures and Their Generated Sentences, in March 2023.

  4. April 2023–2025

    Hardware deployments at Google Japan

    In April 2023, Yamaoka begins working on technical infrastructure at Google Japan. Before starting graduate study in 2025, they spend two and a half years at Google Japan managing hyperscaler network and compute hardware deployments in cross-functional environments.

  5. 2025

    Img2Vocab and MSc in computer science

    In 2025, they co-author the IEEE Access article Img2Vocab: Explore Words Tied to Your Life with LLMs and Social Media Images with Ko Watanabe, Koichi Kise, Andreas Dengel and Shoya Ishimaru. They become an MSc student in computer science in winter 2025, with an interest in statistical machine learning.

  6. December 2025

    Multi-task bandits preprint

    In December 2025, Yamaoka co-authors the preprint Co-Exploration and Co-Exploitation via Shared Structure in Multi-Task Bandits. The paper proposes a Bayesian framework for exploration in contextual multi-task multi-armed bandit settings where user context is partially observed and latent context variables induce dependencies across reward distributions.

  7. February 2026–July 2026

    Linear-LLM-SCM and ICML 2026 workshop

    In February 2026, they post the preprint of Linear-LLM-SCM as first-listed author with seven co-authors, including Eyke Hüllermeier and Sebastian Josef Vollmer, acknowledging funding for the AI4Nof1 project by the state of Rhineland Palatinate. The open-source, plug-and-play framework evaluates how well large language models parametrize linear Gaussian structural causal models when a causal graph is given, and experiments on seven real-world graphs show variability in coefficient estimates. A revised version posted in July 2026 is accepted as a poster at the Structured Data for Health workshop at ICML 2026 in Seoul, South Korea.

  8. As of October 2026

    Research assistant at DFKI

    As of 2026, they are an MSc computer science student and research assistant at DFKI, in the Data Science and its Applications group, with a second affiliation in the Department of Computer Science at the University of Kaiserslautern-Landau (RPTU). Their research on sequential decision-making for personalized medicine covers bandits, preference learning and causality within the AI4Nof1 project.

Career

  1. NowDFKIResearch assistant
  2. Until 2025Google Japan

Education

  1. 2023Osaka Prefecture UniversityBEng
  2. RPTU University Kaiserslautern-LandauMSc

From the biography.

  1. 16% of AI career moves in the 2020s went from a university to a companyIncludes Kanta Yamaoka. We estimate that 16% of AI career moves in the 2020s went from a university to a company; 29% of moves starting at a university went to a company, and 76% ended the university role.
  2. 23% of people shaping AI who left a university for a company later returned to a universityIncludes Kanta Yamaoka. We estimate that 468 of the 2,011 people shaping AI who left a university for a company (23%) later returned to a university after a median of 5 years, and 163 of 829 (20%) made the reverse round trip.
  3. Nonprofit roles are 2.3 times as common in AI safetyIncludes DFKI. We estimate where people focused on safety, alignment and evaluation work. Nonprofit roles appear for 25.1% of safety people with known affiliations, compared with 11.0% in other areas.
  4. The top 10 organizations employed 22% of people shaping AI in 2025: 51% of company researchers and 21% of university researchersIncludes DFKI. We estimate the share of people shaping AI who worked at the top 10 organizations each year from 1955 to 2025, comparing concentration across companies, universities, and institutes and labs.

Colleagues

People named in this biography, and people with dated positions at the same organization in the same years.

Sources and credits

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

  1. LeCycl: Learning Cyclotron opu-imp.github.io. Checked 22 October 2026.
  2. IMP lab. publication database: Detail of Publication publication.imlab.jp. Checked 22 October 2026.
  3. Welcome - Kanta's Blog kanta-yamaoka.earth. Checked 22 October 2026.
  4. Kanta Yamaoka — Data Science and its Applications dsa.dfki.de. Checked 22 October 2026.
  5. Publications - shoya.io shoya.io. Checked 22 October 2026.
  6. [2512.12693] Co-Exploration and Co-Exploitation via Shared Structure in Multi-Task Bandits arxiv.org. Checked 22 October 2026.
  7. [2602.10282] Linear-LLM-SCM: Benchmarking LLMs for Coefficient Elicitation in Linear-Gaussian Causal Models arxiv.org. Checked 22 October 2026.
  8. Linear-LLM-SCM: Benchmarking LLMs for Coefficient Elicitation in Linear-Gaussian Causal Models (arXiv HTML, v2) arxiv.org. Checked 22 October 2026.
  9. Linear-LLM-SCM: Benchmarking LLMs for Coefficient Elicitation in Linear-Gaussian Causal Models (arXiv HTML, v1) arxiv.org. Checked 22 October 2026.
  10. ICML Linear-LLM-SCM: Benchmarking LLMs for Coefficient Elicitation in Linear-Gaussian Causal Models icml.cc. Checked 22 October 2026.