Meta Platforms is establishing a new applied AI engineering organisation to strengthen its superintelligence efforts, according to The Wall Street Journal.
The new teams will be led by Maher Saba, currently a vice-president within the company’s Reality Labs division. The organisation will report to Meta’s Chief Technology Officer, Andrew Bosworth, and is designed to operate with an ultra-flat structure with as many as 50 employees per manager.
The unit will work closely with Meta’s Superintelligence Lab to build what Saba described as “the data engine that helps our models get better, faster”, in an internal memo.
It will comprise two teams: one focused on developing interfaces and tooling, and another responsible for executing tasks, generating data and delivering evaluations that feed back into modelling teams.
“Building great models isn’t just about researchers and compute; it requires real-world data, feedback and evaluations,” Saba wrote. “This creates the flywheel that turns a strong model into a leading one. Recently, we’ve seen strong gains from reinforcement learning and post-training, and we believe we have a genuine opportunity to move faster and pull ahead if we double down on these efforts.”
Meta overhauled its AI strategy last summer, launching the Superintelligence Labs under the leadership of former Scale AI chief Alexandr Wang. His team is tasked with developing the company’s latest AI models.
Since then, Meta has accelerated hiring and is working on new models, codenamed Avocado and Mango. Investors and analysts have been pressing for updates, with chief executive Mark Zuckerberg fielding questions about progress on recent earnings calls.
Speaking in January, Zuckerberg said the company would begin releasing its new models and products in the coming months. “I expect our first models will be good, but more importantly will show the rapid trajectory we’re on,” he said.
The new applied AI engineering unit is recruiting managers from across software engineering, design, product management, data science and data engineering, and has asked interested employees to submit an internal form by Friday.
ALSO READ: The Playground is Closed: 10 Hard Truths from the Cisco AI Summit
