Google DeepMind Engineer's Blunt Job Advice: 'Work Like a Dog'

Google DeepMind distinguished engineer Vladimir Feinberg has a short, direct answer for anyone asking how to land a role at a top AI lab. "Work like a dog," he said. The quote, pulled from a recent discussion on career pathways, cuts through the usual talk of networking and skill-building. At labs like DeepMind, OpenAI, and Anthropic, the bar for entry has climbed steeply in the past two years. Feinberg's advice reflects a reality that hiring data backs up. The number of applicants per open research position has roughly tripled since 2022, based on estimates from multiple industry surveys. Most candidates hold PhDs in machine learning or a related field. Many also have published at top conferences such as NeurIPS or ICML. That baseline alone is no longer enough. Feinberg's point is that the differentiating factor is brute effort on top of strong credentials. The comment resonated widely on professional social networks, where current and former AI researchers echoed the sentiment. One ex-DeepMind researcher said the workload inside the lab is itself a filtering mechanism. "If you can't sustain that pace pre-hire, you won't survive post-hire," the person said. The practical takeaway for anyone aiming at a frontier AI job is that time spent on side projects, open-source contributions, and reproducing results from recent papers may matter as much as formal qualifications. The competitive landscape for AI talent also has read-through for publicly traded companies investing in AI capabilities. Apple, which has been expanding its machine learning hiring, is one example. The same talent pool that labs like DeepMind draw from feeds into Apple's on-device AI efforts. When a distinguished engineer at one of the most selective AI labs says the bar is high and rising, it suggests that the talent shortage is not easing. For companies with large AI ambitions, the cost of attracting and retaining that caliber of engineer will likely stay elevated.