Introducing STATE-Bench: A benchmark for AI agent memory
STATE-Bench is an open-source, memory-agnostic benchmark freely available to agent developers, researchers, and platform teams.
Senior Applied Scientist
Nishant is working on AI agents that can learn from past experience, with a focus on building realistic agent benchmarks. Previously, he worked on agentic content understanding and training custom-LLMs at Microsoft.






Open Source enables Microsoft products and services to bring choice, technology and community to our customers.
Notifications