AutoGen Co-lead
One of the most widely adopted open-source multi-agent frameworks, now the core of Microsoft Agent Framework.
MarkItDown Co-lead
Convert any file to Markdown for LLM pipelines.
Magentic-One Co-lead
Generalist multi-agent system evaluated on GAIA and WebArena for complex web and file tasks.
Magentic-UI Co-lead
Human-centered web agent with co-planning and guardrails.
Magentic Marketplace Co-lead
Collaboration with economists studying what happens when AI agents participate in two-sided markets.
Selected papers. Full list on Google Scholar
Now
Principal Researcher at Microsoft Research AI Frontiers. I co-created AutoGen, one of the most widely adopted open-source frameworks for multi-agent systems, now the foundation of Microsoft's Agent Framework. I then co-led Magentic-One, a generalist system we rigorously evaluated on benchmarks like GAIA and WebArena for complex tasks requiring reasoning, web browsing, and code execution. Building these systems exposed how hard the human side of agents is—we identified twelve fundamental challenges in human-agent communication, from conveying what an agent is about to do, to managing the tension between transparency and information overload. This shaped my subsequent work: Magentic-UI, which I co-led to bring human oversight directly into the agentic loop through co-planning and action guards, and more recently, research on societies of agents—Magentic Marketplace, a collaboration with economists studying what happens when agents participate in two-sided markets at scale.
Before
Ph.D. in Computer Science from University of Washington, advised by Dan Weld. My earlier research focused on human-AI decision making—how AI explanations affect team performance and how to update AI systems without breaking user trust. During my PhD, I interned at Microsoft Research with Besmira Nushi, Ece Kamar, and Eric Horvitz. B.Tech from IIT Delhi, where I worked with Mausam on natural language processing.