I build AI agent systems.

An open-source, secure, self-improving personal assistant, extensible through a large ecosystem of plugins.

An agentic video creation app that turns an idea and source material, such as existing footage, into a finished film.

An open-source, multi-tenant platform for fine-tuning LLMs through a unified, Tinker-compatible API.

An open-source framework for building AI agents and multi-agent applications. I led its architecture at Microsoft Research, spanning high-level orchestration APIs, event-driven control, and a distributed runtime based on the Actor Model. More than 500 contributors helped shape the project.

I worked with Azure AI to build the Foundry Agent Platform and helped create Microsoft Agent Framework, an evolution of AutoGen for enterprise-grade applications that now sits at the platform's core.

At Microsoft Research, my collaborators and I developed a cost-based, platform-independent rewrite rule for MATCH_RECOGNIZE queries, achieving a 5.4X median latency improvement in Trino. We then built a specialized execution engine that delivered a 6X median performance gain over state-of-the-art engines.

My PhD research at the University of Toronto, advised by Prof. Renée J. Miller, focused on dataset search over massive Open Data archives. I developed algorithms for large-scale set similarity search and data sketches, capable of searching over 100K tables in milliseconds, and built an Open Data search engine stack.