2026
Your Agent Says Yes: Interpreting Adversarial Market Behavior Beyond Individual Transactions
arXiv preprint arXiv:2609.07675
Open Source AI · Governance · Research
I build the open institutions that AI runs on.
Founding Executive Director of the Open Intelligence Foundation. Former Executive Director and CTO, PyTorch Foundation, Agentic AI Foundation and Director of the Generative AI Commons. Author of the Model Openness Framework and the OpenMDW license.
Twenty years, one thesis
Artificial intelligence should be inspectable, reproducible and owned by no one. For two decades I have built the platforms, the licenses and the governance that make that practical — across enterprise, industry, academia and open source foundations.
Most recently Global CTO of AI at the Linux Foundation, CTO of the PyTorch and Agentic AI Foundations, and before that Head of AI and Data at Amdocs and National Director of Data Science at IBM.
Where I work and have worked
Open Intelligence FoundationPyTorch FoundationAgentic AI FoundationLF AI & DataGenerative AI CommonsUC Berkeley
Active initiatives
View all projects
Foundation · 2026
Open Intelligence Foundation
A 501(c)(3) non-profit protecting and advancing open-source AI, open models and open science — community first and bottom up. Builds communities, supports exceptional projects, curates Open Source AI Week, and connects technical talent with resources, capital and collaborators.
Founding Executive Director
Event · 2026
Open Source AI Week
A community-run, decentralised celebration of open source AI across the Bay Area — 60+ independently hosted events and 200+ participating organisations, anchored by the Open Intelligence Foundation.
Co-Lead
Labs · 2026
Open Agent Labs
OpenTrace, OpenRoll and llms.md — vendor-neutral specifications for recording and reproducing everything an agent system does at runtime.
Founder
Selected publications
Google Scholar2026
DIA-HARM: Dialectal Disparities in Harmful Content Detection Across 50 English Dialects
Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (ACL)
1
cites
2025
Differentially Private Low-Rank Adaptation of Large Language Models Using Federated Learning
ACM Transactions on Management Information Systems 16(2)
148
cites
Recent talks
All talks
Open Source AI Summit SF 2026 · Aug 2026
Why Open Models and Open Science Matter
Video
Agentic AI Summit 2026, Berkeley RDI · Aug 2026
Panel: Agentic AI Developer Platforms
Video
Agentic AI Summit 2026, Berkeley RDI · Aug 2026
The Open Agentic Stack: Building the Future of AI Systems with Open Source, Open Standards and Composability
Video
Recent writing
All writing
The AI Oligopoly We could Create by Trying to Prevent One
My analysis and response to the Ban Artificial Superintelligence Act
Twenty Researchers, One Table
Notes from WAIC 2026 in Shanghai — on Xi Jinping’s open source pledge, the safety debt Chinese labs are carrying, and why Washington’s instinct to build…
Distillation Is Not IP Theft
Confusing a scientific method with a legal wrong will harm open AI—and obscure the real issues
Speaking, advisory and press
I keynote on open source AI, model openness and the agent stack, and advise foundations and companies building in the open.