Overview
Consistent with a career centered around building technology that gives people agency through open source and community building, my last three years have been focused on the intersection of technology, policy, and public interest to shape how AI develops, and who it ultimately serves. My projects range from collaborating with startups and companies on responsible AI strategies, to building shared norms and frameworks for openness across the entire AI stack: datasets, models, licensing, and infrastructure. In partnership with Mozilla, Columbia University and like EleutherAI, we convened leading researchers and practitioners to create community-driven tools and practices for open datasets, while also incubating products designed for the public good that translate research into usable technologies. Our research extended into AI for education, exploring how generative systems transform classrooms and learning, and into AI hardware, where open, transparent infrastructure is becoming as critical as open data. Across these efforts, the throughline has been collaboration — bringing together engineers, legal experts, educators, and civil society to design AI that is not only powerful, but also transparent, equitable, and accountable.
Press
- AI Hardware Must Be Open, MIT Tech Review, June 2025
- We finally have a definition for open-source AI, MIT Tech Review, Aug 2024
Publications
- AI Hardware Must Be Open, MIT Tech Review, June 2025
- Accelerating Progress Towards Trustworthy AI, Mozilla 2024
- A different take on AI safety, Arxiv, 2025
- Open Datasets for LLM Training, Arxiv, 2025
- A different take on AI safety, Arxiv, 2025
- Open Datasets for LLM Training, Arxiv 2024
- Towards a Framework for Openness in Foundation Models, Arxiv, 2024