Zhiqing Sun

I am a Research Lead at Meta TBD Labs, part of Meta Superintelligence Labs, where I founded and lead the agent research team.

At Meta, I oversee agent research for Muse Spark 1.0–1.3 and the modeling effort behind Muse (Hatch), our recently launched personal AI agent powered by Muse Spark 1.3. Built to bring personal AI agents to everyone, Muse is now live. Give it a try and let me know what you think.

Previously, I was a Research Scientist at OpenAI, where I worked on frontier post-training for reasoning and agents. I was research co-lead for Deep Research and ChatGPT Agent, and contributed to o3/o4-mini and Operator.

I received my Ph.D. in Computer Science from Carnegie Mellon University, advised by Yiming Yang, where I worked on scalable alignment, reasoning, and self-improving language models. Before that, I received my B.S. with honors in Computer Science from Peking University.

Email / Google Scholar / X / GitHub

Selected Work

• Sept 2026 Muse Agent — Modeling / research lead
  End-to-end ownership of model training and agent capabilities.
• Apr–Sept 2026 Muse Spark 1.0 · 1.1 · 1.2 · 1.3 — Agents lead
• July 2025 ChatGPT Agent — Research co-lead
  Launch livestream (w/ Sam Altman, Casey Chu, Isa Fulford, and Yash Kumar)
• Apr 2025 OpenAI o3 / o4-mini — Contributor
• Feb 2025 OpenAI Deep Research Agent — Research co-lead
  OpenAI Forum livestream (w/ Isa Fulford)
• Jan 2025 OpenAI Computer-Using Agent (Operator) — Contributor

Selected Papers

• BrowseComp: A Simple Yet Challenging Benchmark for Browsing Agents OpenAI, 2025 · Equal contribution
• Easy-to-Hard Generalization: Scalable Alignment Beyond Human Supervision NeurIPS 2024
• SALMON: Self-Alignment with Instructable Reward Models ICLR 2024
• Principle-Driven Self-Alignment of Language Models from Scratch with Minimal Human Supervision NeurIPS 2023 · Spotlight
• DIFUSCO: Graph-based Diffusion Solvers for Combinatorial Optimization NeurIPS 2023 · Spotlight
• MobileBERT: a Compact Task-Agnostic BERT for Resource-Limited Devices ACL 2020
• RotatE: Knowledge Graph Embedding by Relational Rotation in Complex Space ICLR 2019

Full publication list on Google Scholar.

Ph.D. thesis: Scalable Alignment of Large Language Models Towards Truth Seeking, Complex Reasoning, and Human Values.