Google DeepMind Research Scientist Interview Questions
& Process

I had a referral. Went through an initial HR screening, then a technical quiz, followed by a coding interview. After that, I interviewed with a researcher. Unfortunately, I got a rejection email stating that my research interests didn't align with their current priorities, but they'd keep me in mind. The interview with the researcher was a bit unclear in its purpose, with questions that seemed trivial or vague, possibly gauging my understanding of the topic or my fit with their research. They might have been assessing my fit with the group's research interests, but I don't think the confusion was intentional. If I had passed, the next steps would've included a one-hour candidate talk, about five more interviews with researchers, and a final 'people and culture' interview. They provide preparation tips at each stage, which were helpful for the technical quiz and coding interview.

I found out about a research scientist opening that matched my interests. I was a bit hesitant about the company, but their tech resources (Borg) were appealing. The interviews spanned two months, mostly due to my schedule. It started with a brief HR call about my work style and reasons for wanting to join DM. Then came the technical part: two rounds with questions that felt pretty basic, covering high-school level math, algorithms, stats, and ML. The interview style was odd; it felt like a quiz, with the interviewer being unresponsive when I tried to converse. I'd prefer a more conversational format to explain my answers and assess potential colleagues. After that, there was a programming interview. It involved simple practical questions and a coding exercise in a shared Google doc, using any language. I primarily use C++, but I'd heard they favor Python, so I quickly brushed up and felt prepared. The interviewer was pleasant and we had a casual chat at the end. The final interview was with three team members I'd be working with. HR gave me their names beforehand so I could look into their work. I was surprised by their research statements, which seemed overly ambitious for simple AI models and analyses, often presented as solutions to complex problems. The first interviewer was late, unapologetic, and seemed disengaged, assuming I knew who he was. I tried discussing their publications, but he seemed unfamiliar with them. When I brought up a math problem in their methodology, he dismissed it. The second interviewer also assumed I knew him, and when I joked about their niceness, he became condescending, stating DM, as a 'money-making' entity like Google, had no room for my 'pro-social' research interests. He then boasted about their accomplishments and persistently questioned my career 'changes.' I just followed my interests and opportunities. He also asked a few other brief questions. I had one more brief meeting, mainly for the person who referred me. I met a research engineer who was quite arrogant. He introduced himself, again assuming I should know him. He asked about my work in a way that implied familiarity, which felt unprofessional given my diverse background. He asked for details on my recent project techniques and my plans for DM, focusing on a simplified explanation of drawing from a distribution rather than deeper statistical or coding challenges. I later discussed the interview with my referrer, and their experience was completely different. I wondered why, given our similar backgrounds, skills, and personalities, except for gender – I'm a woman, and they hired few women. I suspect my CV was flagged by an AI, and the interviewers had to comply. Later, HR informed me they weren't moving forward. I thanked them, disappointed but also relieved. Regardless of sexism, their research seemed 'overrated,' and the company culture focused on quantity over quality and self-congratulation would have been tough. It might be a good fit for new PhDs or aspiring academics seeking better pay than academia without the rigor of a typical business, but the environment seemed unbearable to me. I hope their biased statistical research doesn't impact my life.

I went through a 2-part interview process. The first part focused on coding and computer science fundamentals, like algorithms, space and time complexity, and concepts such as mutexes and semaphores. The second part was about statistics, probability, and understanding code. They asked about binomial distributions and specifics of certain neural network architectures, like CNNs, which I didn't have prior experience with. Honestly, it felt harder than it needed to be and was very specialized. The first interviewer was nice, but the second one was totally unprofessional and not suited for interviewing, which really put me off the company.