Matthew Kenney

I'm the founder of Algorithmic Research Group, where I lead research on benchmarks, environments, and multi-agent systems for understanding recursive self-improvement.

I've collaborated with researchers from OpenAI, Anthropic, and DeepMind on advancing the state of the art in language models and AI safety. My technical work spans transformer pre-training, some minimal post-training/RLing, model quantization, RAG systems, and agents. One might call me a generalist.

Previously, I was a Senior ML Engineer at Apple and at Alethea building models and applications. Before that, I was an Assistant Research Professor at Duke University. I served as a Visiting Researcher at Constellation. I currently teach a class at Duke on the side. I used to be a new media artist building weird stuff with code.

  • DESIGNTK 531: AI-Integrated Hardware Prototyping and Interaction Design — Duke University, Pratt School of Engineering, MEng in Design & Technology Innovation (2025–Present)
  • EGR 190/EGR 590: AI for Everyone — Duke University (Spring 2020)
  • EGR 190-02/EGR 590-02: Machine Learning Methods and Practice — Duke University (Fall 2019)
  • ISS 294L: Interactive Graphics — Duke University (Spring 2018, Spring 2019)
  • CMAC 564S: Physical Computing — Duke University (Fall 2018)
  • MFAEDA 713-0: Computational Media — Duke University, Co-Instructor (Fall 2018)
  • Algorithmic Research Group — Open Philanthropy (2024–Present)
  • Agentic AI R&D Grant — Open Philanthropy (2023–2024)
  • AI Agents for AI Research and Development — Constellation Visiting Researcher Forum, Berkeley, CA (2023)
  • Big Problems in Big Tech — Duke University Annual Provost's Forum (2023)
  • Disinformation in the Age of AI — Keynote, International Conference on Machine Learning (ICML) BIAI track, Baltimore, MD (2022)
  • Creative Red Teaming: Approaches to Addressing Bias and Misuse in Machine Learning — Keynote, Conference on Neural Information Processing (NeurIPS) BIAI track, Vancouver, Canada (2020)
  • AI Index Search Engine — Stanford Human-Centered AI Forum, Palo Alto, CA (2019)
  • At the Intersection of AI and Disinformation — Duke University Annual Provost's Forum, Durham, NC (2019)
  • Deepfakes: Commodification, Consequences, and Countermeasures — Open Data Science Conference West, San Francisco, CA (2019)
All creative work →