Google DeepMind: Key Tips for Landing an AI Job

Advice from a Google DeepMind Engineer for Succeeding in AI

Vladimir Feinberg, a renowned engineer at Google DeepMind, recently shared his recommendations for those aspiring to join a leading artificial intelligence laboratory. According to him, three qualities are essential: intention, mathematical maturity, and perseverance. Feinberg advises future AI researchers to engage in demanding, evidence-based courses while developing their programming skills.

A Traditional Approach for a Competitive Sector

Feinberg emphasizes that landing a job at a top AI lab may require a relatively traditional method. "Work like a madman," he writes in his blog titled "How to Land a Job at a Leading Lab." The competition is fierce, making access to these positions particularly challenging.

He explains that there exists an elite group of students, both at the undergraduate and doctoral levels, who stand out through their research in machine learning at prestigious conferences, their participation in math and programming competitions, and their connections with these labs established through older classmates or friends.

A Shift in Target for Talent

Feinberg notes that a few years ago, this same group of talent was often recruited by Wall Street giants like Citadel and Jane Street. Today, these high-achieving students are increasingly turning to renowned AI companies like OpenAI, Anthropic, and Google DeepMind.

According to Feinberg, these students succeed because their traits are highly predictive of success. He therefore encourages aspiring AI researchers to join this cohort by investing in challenging courses, coding intensively, and proactively using AI tools.

The Importance of Mathematical Maturity

For Feinberg, it is crucial to dedicate long hours outside of classes, including nights and weekends, to develop the skills necessary to compete for jobs in leading AI labs. "There is no substitute for this to achieve mathematical maturity, which is essential," he asserts. However, he adds that demonstrating a specific skill required by a lab is also an obvious way to get hired.

Working at the Frontiers of AI

Feinberg admits that breaking into an AI lab can seem like a vicious cycle. To overcome this, he recommends working at the frontiers of what leading labs are doing, particularly in the creation of large language models (LLMs). This involves understanding what these models need to function and identifying the touchpoints for their outcomes. These specific areas, while not requiring training in LLMs, are nonetheless essential for the enterprise.

General Career Advice

In addition to his technical advice, Feinberg shares a general career tip: "Be the kind of colleague that people want to see succeed." He suggests identifying opportunities for the team's complementary skills to shine, crediting collaborators for their leadership, and choosing projects where your success contributes to that of others.

Reactions and Perspectives

In an episode of the Peterman Pod published on June 15, Feinberg mentioned that his blog post received positive feedback from people working at Anthropic and OpenAI, who endorsed his advice.

While he acknowledges that the business strategies and offerings of labs may vary based on their specialties and clients, Feinberg believes there are many commonalities among labs regarding the skills they seek.

When asked about the possibility that advancements in AI might diminish the value of research work, Feinberg remains optimistic. He thinks that research skills will become increasingly crucial. "Thinking about how to build systems around these LLMs to do my job more efficiently — that's what will set you apart in the future," he asserts. "And I believe that's true no matter what you end up doing."