Is Meta Breaking Its Engineering Organization

Meta has undertaken a massive organizational overhaul to secure leadership in the AI platform race, acquiring a 49% stake in Scale AI for $14.8 billion and appointing Alexandr Wang as Chief AI Strategy Officer.

In this process, 30-50% of the existing product engineering team has been forcibly moved to the ADO (Agent Data Optimisation) organization, and an estimated 4,000-5,000 software engineers are being deployed to data labeling and RLHF (Reinforcement Learning from Human Feedback) work.

The major changes and issues are as follows:

• Data-centric operations: Meta introduced a system that tracks engineers' keystrokes and mouse clicks to generate AI training data, but due to privacy invasion controversy, it is being operated on a limited basis in some regions (such as the UK).
• Culture clash: As the existing engineering culture, which valued autonomy and impact, is being replaced by top-down data optimization work, internal backlash is intensifying, particularly among infrastructure and security teams.
• Performance pressure: Along with managing token usage to optimize AI utilization, anxiety within the organization is escalating due to the recent announcement of a 10%-scale workforce reduction.

As a result, criticism has emerged that Meta's engineering organization is being treated like a Cost Center that produces data for AI training, rather than a Profit Center that drives innovation.