Jason Wei Leaves Meta to Join Mercor as Chief Scientist
Jason Wei, first author of the chain-of-thought paper and a former OpenAI o1 contributor, has confirmed his departure from Meta's TBD lab to become Chief Scientist at AI data company Mercor, where he will build a research team focused on data and evaluation. The move underscores data and evals as the next competitive frontier and puts Meta, which paid $14.3B for a 49% stake in Scale AI, in an awkward spot.
Full Report
宝玉 (@dotey) · 2026-11-18 00
I didn't expect that Jason Wei had already left Meta to become Chief Scientist at the AI data company Mercor.
On November 17, US time, Mercor CEO Brendan Foody announced that Jason Wei would serve as Chief Scientist, responsible for building a research team and overseeing data and evaluation. Jason Wei subsequently posted to confirm: after spending a year at Meta's TBD lab, "there are more companies training AI models than ever before, and data will determine the next frontier."
At Google Brain, Jason Wei was the first author of the chain-of-thought paper, which had models write out reasoning steps before giving an answer—an approach that was later widely adopted. He then worked at OpenAI on o1 and Deep Research, and in July 2025 was poached by Meta for a hefty sum.
His other strength is creating "exam questions" for AI. SimpleQA, which he led, tests whether models make things up, and BrowseComp tests whether AI can scour the web to find obscure answers; both have become commonly used evaluations when companies release models, and he also participated in HealthBench, which uses doctors for scoring. In his blog, he wrote a "verifier's law": the easier a task is to verify, the easier it is to train AI to learn it.
That is exactly the business Mercor is in: its platform has more than 5 million doctors, lawyers, and engineers who help AI companies create questions, write reference answers, and score models. On November 10, it had just raised $750 million at a $20 billion valuation, doubling in a year. Now, not only major labs but also a growing number of enterprises are using open-source models to train their own, and they all need this kind of data and evaluation. Mercor began building a research team in October, and Edward Hu, an author of LoRA, has already joined.
This is a bit awkward for Meta. The TBD lab is led by Scale AI founder Alexandr Wang; Meta spent $14.3 billion in 2025 to buy 49% of Scale's shares, and Mercor is Scale's main competitor. TechCrunch reported last year that TBD researchers found Scale's data quality poor and preferred to use Mercor's and Surge's. This is also another prominent researcher to leave TBD after Andrew Tulloch was reported in September to have moved to Anthropic.
Jason Wei did not say why he left Meta, and Meta has not yet responded.
↳ Quoting @_jasonwei (Jason Wei): After an amazing year at Meta’s TBD lab, I am very excited to join Mercor as Chief Scientist. Data will shape the next frontier as more companies train AI models than ever before. Mercor is incredible and I look forward to building the future with Brendan, Adarsh, and team!