Meta Restructures AI Team for Speed
Meta has created a new Applied AI Engineering division with an extremely flat structure, aiming to 'fast-track superintelligence development.' Under CTO Andrew Bosworth, teams may have up to 50 engineers per manager, a move that underscores a new emphasis on autonomous, self-directed engineers.
This new Applied AI Engineering group is led by Maher Saba, a VP from Meta's Reality Labs, and will work in partnership with the Meta Superintelligence Lab. That lab is headed by Alexandr Wang, the former CEO of Scale AI, who joined Meta after it invested $14.3 billion for a 49% stake in his company. The division's specific mandate is to build the "data engine that helps our models get better, faster," according to an internal memo. It consists of two teams: one building interfaces and tools, and another focused on data collection and evaluation to create the feedback loop that turns a "strong model into a leading one." This move is one of at least four major AI-related restructurings at Meta in the past year, signaling intense urgency to streamline development. Previous shakeups have aimed to reduce overlap between research and product teams and address internal concerns about leadership and the rapid pace of development. The flat management model, sometimes with only one manager for every 50 engineers, is a deliberate strategy to increase speed by empowering individual engineers with more autonomy. The structure is popular with startups to reduce bureaucracy and accelerate decision-making, a culture CEO Mark Zuckerberg wants to foster as Meta competes with OpenAI and Google DeepMind. This reorganization is backed by massive capital investment, with Meta planning to spend between $115 and $135 billion on AI infrastructure in 2026. The company has also secured a $100 billion deal with AMD for custom AI chips to power its data centers.
