Da Yin is a Postdoctoral Researcher at Metas FAIR Supertelligence Lab, specializing in adaptive and self-evolving LLM agents. A PhD graduate from UCLA and a 2023 Amazon PhD Fellow, his research focuses on agent learning and post-training. His work includes highly-cited open-source projects like Agent Lumos and ETO.
His research on Agent Lumos aims to build transparent, open-source alternatives to closed-source AI, directly addressing issues of reproducibility and affordability in the field.
Da Yin
Postdoctoral Researcher at Meta
Redmond, Washington, United States
Overview
Topics They Care About
Self-Evolving Agents
His core research focus at Meta and during his PhD is on building adaptive, self-evolving LLM agents that can learn and improve on their own.
Open-Source AI
He co-authored the Agent Lumos framework, explicitly created to be an open-source alternative to closed, black-box agents from major labs.
AI Trajectory Optimization
His recent work, ETO, presented at ACL 2024, introduces a method for LLM agents to learn from failures, inspired by human "trial and error".
Media Appearances
Da has no verified media appearances
Work History
5-2025
Postdoctoral Researcher at Meta
Education
9-2020 - 4-2025
Doctor of Philosophy - PhD from UCLA
9-2016 - 7-2020
Bachelor's degree from Peking University
More Information
Social Presence :
Prographics :
Exp : 1
Location : Redmond, Washington, United States
Job Level : N/A
Designation : Postdoctoral Researcher at Meta
