Meta is putting billions in its newly formed superintending lab, which hits some of the best minds in artificial intelligence. But a few months after its construction, things have already started moving within the team. Several high-profile researchers are moving away, including AI scientist Rishabh Aggarwal, who joined the technical veteran with a million-dollar salary in April. After just five months, he has now announced his exit to follow “a different type of risk”.

Aggarwal announced his departure from MSL in a post on X. “This is my last week on @Aaitmeta. It was a difficult decision that the new supervision is not released with TBD lab, especially given talent and calculation density.

While Aggarwal has confirmed his exit from the Meta Superintending Lab, his next step is not clear. He has not revealed that he is joining another company, starting something on his own, or chasing educational research.

But Agrawal is not the only high-profile exit from Meta’s AI lab. A wired report shows that at least three researchers have left the Superintendent team in recent weeks. Two of them have already returned to openi. There is an AV Verma, who was also a researcher in Openai before joining the first meta. There is another Ethan Night, who had previously worked in Openi, then Aleon moved to Musk’s XE, and later joined the meta before returning to Openai.

All three exit comes from the middle of the Elite Higher Meta made as part of their ambitious push to create a “superintendent” earlier this year, a program was announced by Zuckerberg just two months ago. The new recruits were reportedly offered a package of compensation packages of multi-mill-dollar, in which the company aggressively targets talent from rivals such as Google Deepmind, Openai and XAI.

Rishabh Aggarwal’s work in Meta

Meta hired Aggarwal on a million-dollar salary package. During his short tenure, Aggarwal revealed that he worked after training for the idea model, which includes scales of reinforcement learning methods and improving distillation techniques.

Meanwhile, Meta spokesperson Dave Arnold reduced the resignation, said, “During an intensive recruitment process, some people will decide to stay in their current job instead of starting a new. It is normal.”

Who is Rishabh Aggarwal?

Academically, Aggarwal is an alumnus in IIT Bombay, where he studied computer science and engineering, before he got a PhD at Cubek Artificial Intelligence Institute. In his early research, Aggarwal discovered learning and its applications, including innovative assessment techniques.

Following his educational work, Aggarwal started his career with an internship at Saavn, Tower Research Capital and Waymo before joining Google Brain as a senior research scientist in 2018. Over the next five years, he contributed significantly to learning deep reinforcement, including research that won the prestigious Neurips 2021 Best Paper Award.

Aggarwal later moved to Deepmind of Google, where he focused on advancing large language models using reinforcement learning and self-reforming methods. In April 2024, he was admitted by the Superintendent Lab of Meta. While his next step is not yet known, Aggarwal has also been active in academia, serving as an assistant professor at McGill University.