Meta shifts engineers to Applied AI
Meta has begun compelling selected top engineers to transfer into a new Applied AI Engineering unit, a move that suggests the company prioritises turning AI capability into shipped systems rather than letting transfers be purely voluntary. The Times of India reports the change as a compulsory redeployment for some engineers (timesofindia.indiatimes.com).
Meta has started telling selected software engineers they are moving into a new Applied AI Engineering unit, and Reuters reported the transfer is no longer being treated as optional for everyone chosen. The internal memo said notices began going out this week as Meta reorganized teams ahead of layoffs. (reuters.com) This is not Meta hiring a few more artificial intelligence specialists on the side. It is Meta pulling proven engineers out of other parts of the company and putting them into one group built to turn artificial intelligence into tools that ship inside Meta. (reuters.com) The new group is called Applied AI Engineering, and it is led by Maher Saba, a vice president from Reality Labs, the division that builds Meta’s virtual reality and hardware products. Reuters said Saba first opened the team to volunteers in March before the company switched to directed transfers. (reuters.com) Inside Meta, “applied” means using models to do work, not just training bigger models in a lab. Reuters reported the unit is supposed to build artificial intelligence agents that could eventually handle a large share of Meta’s own engineering work. (reuters.com) That fits what Mark Zuckerberg has been saying in public since January. In an interview with podcaster Joe Rogan, Zuckerberg said Meta expected artificial intelligence in 2025 to be able to act like a midlevel engineer that can write code inside the company. (reuters.com) Meta has been building the raw ingredients for that push for more than a year. On April 5, 2025, the company unveiled Llama 4, a new generation of its large language models, and on April 29, 2025, it launched a standalone Meta AI app built on Llama 4. (about.fb.com) A large language model is the prediction engine underneath a chatbot, like the motor under a car hood. An applied engineering team is the part that turns that motor into a delivery van, a taxi, or a factory machine people actually use every day. (about.fb.com) (reuters.com) Meta is also reshaping the company around this bet in other ways. Reuters said the transfers are part of a broader workforce reorganization tied to planned layoffs, which means the artificial intelligence build-out is happening alongside cuts elsewhere. (reuters.com) The pressure is easy to see from the outside. Meta has more than 3 billion people using apps across Facebook, Instagram, WhatsApp, and Messenger, so even a small artificial intelligence tool for coding, moderation, ads, or customer support can touch systems at enormous scale. (theverge.com) So the signal in this move is not just that Meta wants better artificial intelligence models. It is that Meta wants its best builders working on the part that turns artificial intelligence into internal software, and it is willing to reassign people directly to get there faster. (reuters.com)