Behind the Scenes: A Meta GenAI Engineer Speaks Out Amid Workforce Cuts
As Meta Platforms prepares for its next round of layoffs—impacting approximately 3,600 employees—concerns over leadership, organizational structure, and workplace culture are growing. CEO Mark Zuckerberg’s push for a leaner, high-performance workforce has triggered anxiety across departments, particularly in high-pressure divisions like GenAI.
To shed light on the internal reality, we spoke with a Meta GenAI engineer today, who requested anonymity, about their experience in one of the company’s most demanding teams. In this exclusive interview, he reveals the intense work culture, chaotic leadership, and mounting competition from rivals like DeepSeek. From grueling hours to the lack of a clear strategic vision, his candid insights paint a stark picture of life inside Meta’s AI division.
As Meta doubles down on efficiency and output, this interview provides a rare glimpse into the company’s internal struggles and the growing disillusionment among its top engineers. Is Meta’s high-stakes AI race sustainable, or is it pushing its workforce to the breaking point? Will it win in the end? Read on to find out.
Exclusive Interview with Meta GenAI Engineer: Inside the Challenges of Working in a High-Stress Environment
Xia: Thank you for agreeing to this interview. Could you share your experience working in Meta's GenAI division?
Dev437: It's been intense. The workload is grueling. We’re expected to be available from 6am ET until 11pm or later. The East Coast team starts early, the West Coast follows suit, and our European colleagues align with the West Coast hours. Even on holidays, there’s rarely any downtime—our message threads are relentless.
Xia: What does a typical workday look like for you?
Dev437: Exhausting. Emergencies often arise over weekends—one of my colleagues worked until 4am just last weekend. The roadmap shifts pretty much every quarter, and we find ourselves scrambling to catch up with the latest technologies that emerge.
Xia: I can sense some frustration. The phrase "incompetent general exhausts his army" comes to mind. Would you say that describes your situation?
Dev437: Absolutely, though it’s more nuanced than that. The reality is, it’s like a Hunger Games scenario. The managers in the middle are in fierce competition to secure project scopes. The leadership at the top doesn’t have a clear vision but keeps pushing for more output and results. We're not afraid of hard work—we deliver. But the organization is incredibly top-down yet the top leadership has no vision, and I’m not sure how long this model can hold up.
Xia: I’ve heard there were some significant organizational changes this week. How is that impacting the team?
Dev437: Yes, there’s been a reorganization, and decision-making power has become more centralized. The direction is constantly shifting, and there's still this relentless pressure for more output with no clear vision. As I mentioned, it's exhausting.
Xia: You mentioned competition earlier. How do you think Meta’s GenAI performance stacks up against competitors?
Dev437: That’s the most frustrating part. On paper, we have the best resources. Our team’s background is stronger than that of many competitors, Google or Deepseek. But despite this, our LLaMA model is significantly behind DeepSeek. The issue? Leadership, in my opinion. It’s a clear disconnect.
Xia: Could you go into more detail about the management structure at Meta?
Dev437: We have a lot of V-teams, each with 7-8 team leads managing relatively small engineering teams—think two-pizza-sized teams. But the org is overly complex, and the targets keep changing. There are weekly sprints, constant updates, and endless progress checks. It’s all a massive communication burden that ends up being counterproductive.
Xia: With such a demanding environment, what’s the work-life balance like at Meta?
Dev437: During the hiring process, they warned us this wouldn’t be your typical work-life balance role. They intentionally filtered out people who cared about work-life balance. Even if you’re in the top 10% of engineers, you need to be in the top 5% to survive comfortably here. Some engineers who were among the best have already sought internal transfers due to the stress. That should tell you a lot.
Xia: Let’s shift gears a bit. How do you view the threat from DeepSeek? What is Meta’s strategic approach to dealing with it?
Dev437: It’s ironic, really. Zuckerberg’s move with LLaMA was essentially a countermeasure to OpenAI’s GPT. It was like, “Oh, you’ve got a fancy, closed-source GPT, OpenAI? Well, I’m going to release an open-sourced, free model so you can’t make money off of GPT either.” But then, DeepSeek came along, and it blew us out of the water. Ironically, though, DeepSeek’s success has actually helped Meta’s stock rise, because investors believe DeepSeek’s technology will eventually benefit Meta. It’s a strange world we live in.
Xia: Speaking of Zuckerberg, he’s been known to warn employees about exposing information online, like on Teamblind. Do you have any concerns about talking to us?
Dev437: No, as long as my identity remains anonymous, I’m not concerned. We’re all pretty used to his management style by now. For example, think about how he switched his public persona in relation to Trump. It's part of the course at Meta.
Xia: What are your future plans? Do you have any thoughts on other opportunities outside Meta?
Dev437: Honestly, there’s no team at Meta I would want to transfer to, and I’m not fond of the GenAI organization. Right now, I’m just waiting for my PSC (performance review) results, and I’m likely going to leave afterward. And by the way, Meta’s recruiters seem to be desperate at the moment. They’re the only ones still aggressively recruiting in this market downturn.
Xia: To wrap up, what are your predictions for Meta’s future? Any unique insights?
Dev437: Interesting question. I haven’t really discussed this with anyone before, but honestly, it all feels like a staged performance. Everyone is forced to appear as if they’re working insanely hard—really, really hard—just to keep investors convinced. But in reality, most of our efforts haven’t led to anything truly groundbreaking. Meta’s earnings remain stable because it dominates digital marketing, and as long as it’s the only company outside of China capable of building open-source LLMs, it’ll be fine. But if you ask me about real innovation, I can’t think of any major product that has captured the moment in a long time. Maybe a few developer tools or open-source LLMs, but those don’t even generate significant revenue.