Researcher

Eric Mibuari

Eric Mibuari is a computer scientist and AI safety researcher whose work spans community technology in Kenya, AI-mediated markets, and safer AI systems. He founded a computer center in Laare while studying aeronautics and astronautics at MIT, then coauthored applied-computing projects on rural health access, interactive education, and Nairobi traffic. At Harvard, he pursued computer science research on African language technology. His 2022 paper examined AI collusion in online markets, and his 2025 work covered clinical language models for Kenyan care and African disease representation in medical benchmarks. As of October 2026, he is an AI Safety Research Scientist at Partnership on AI, where he coauthored a report on monitoring AI agents.

Timeline

  1. 2005–2006

    A computer center in Laare

    In 2005, he founds the Laare Community Technology Centre with the MIT Public Service Center. In 2006, as a graduating senior in aeronautics and astronautics at MIT, he returns to Laare for management seminars and a student-recruitment plan.

  2. 2011–2014

    Computing for health, school, and traffic

    From 2011 to 2014, he coauthors work on prenatal information access in rural Laare, interactive secondary-school lessons, and traffic management in Nairobi. In 2014, the Twende-Twende app combines low-cost camera images and network-flow algorithms with local landmarks to provide drivers with real-time traffic information and suggested routes.

  3. 2021

    Language and technology in Africa

    In 2021, he is a Harvard computer science Ph.D. student and collaborates with John Mugane, director of the African Language Program in Harvard’s Faculty of Arts and Sciences, on “Language, Culture and Technology in Africa.”

  4. 2022

    AI collusion on platforms

    In 2022, he coauthors “Learning to Mitigate AI Collusion on Economic Platforms.” The paper tests whether platform “buy box” rules can curb tacit collusion by reinforcement-learning sellers while maintaining consumer welfare.

  5. 2025

    Clinical language models in Kenya

    In 2025, he coauthors a method for testing language models in Kenyan primary care against Kenya’s national clinical guidelines. He also coauthors a study finding that medical LLM benchmarks underrepresent African disease burdens, which risks misleading performance claims.

  6. As of October 2026

    AI safety at Partnership on AI

    As of October 2026, he is an AI Safety Research Scientist at Partnership on AI. In October 2026, he coauthors The Observability Gap in AI Agents, which identifies six gaps in agent telemetry and calls for coordinated action across framework developers, model providers, enterprises, and regulators.

Sources and credits

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

  1. Kenyan grad shares what he's learned news.mit.edu. Checked 19 October 2026.
  2. Imara Project: Making a Difference csail.mit.edu. Checked 19 October 2026.
  3. Supporting greater access to pre- and post-natal information and services for women in Rural Kenya for CHI 2011 research.ibm.com. Checked 19 October 2026.
  4. Leveraging raspberry pi for interactive education for ACM DEV 2013 research.ibm.com. Checked 19 October 2026.
  5. Twende-Twende: A mobile application for traffic congestion awareness and routing for MOBILESoft 2014 research.ibm.com. Checked 19 October 2026.
  6. Awardees announced for the Motsepe Presidential Research Accelerator Fund for Africa news.harvard.edu. Checked 19 October 2026.
  7. Learning to Mitigate AI Collusion on Economic Platforms arxiv.org. Checked 19 October 2026.
  8. Retrieval-Augmented Clinical Benchmarking for Contextual Model Testing in Kenyan Primary Care: A Methodology Paper arxiv.org. Checked 19 October 2026.
  9. Mind the Gap: Evaluating the Representativeness of Quantitative Medical Language Reasoning LLM Benchmarks for African Disease Burdens arxiv.org. Checked 19 October 2026.
  10. The Observability Gap in AI Agents partnershiponai.org. Checked 19 October 2026.
  11. Our Team - Partnership on AI partnershiponai.org. Checked 19 October 2026.