Google DeepMind Interview Questions
& Process

I applied in late December for this role, initially full-time but then it was changed to a 12-month FTC. The whole thing took 6 weeks and had four stages: first a TA screen, then a technical chat with the Hiring Manager about automation architecture, then a cross-functional interview, and finally a strategic round with a Director. The interviewers were great and asked a mix of technical case studies and big-picture behavioral questions, really making me think. But the follow-up was awful. After the last interview, I heard nothing for two weeks until I reached out to the TA. They gave a weak reason about 'changing the job description' and said they'd update me in a week. It's now been over a month with no final decision or any feedback. The team seems amazing, but the recruiting process and communication were super disappointing. I always wanted to work at DeepMind, but this experience has really soured my view of the company.

Interviewing with Google DeepMind is like entering a focused space where questions aim to understand your thinking process. It typically starts with a recruiter call, which is straightforward and sets the tone. Then, the actual interviews begin with technical screens. These aren't just about solving problems, but more about explaining your thought process as you work through them. Expect a lot of whiteboarding and discussions about your gut feelings and the 'why' behind your actions. If you get through those, the onsite interviews are like a deep thinking marathon, covering research questions, system design, conceptual thinking, and handling uncertainty with open-ended prompts. The interviewers are intelligent and approachable, challenging your ideas constructively. It feels more like a collaborative discussion than a strict test, aiming to see how you think and imagine, not just what you can create. The whole experience is intellectually demanding and thorough, but also respectful, leaving you feeling mentally sharpened.

Got in with a referral. The process was super slow, often took over a month to hear back after interviews. The questions weren't insane, but covered a lot of ground with a high bar. Programming questions were unexpectedly easy. My system design interview didn't go well, which is why I got rejected. It was a really open-ended question with very little time, kind of like a software system design interview which I wasn't familiar with. I had a robotics quiz covering various topics like kinematics, filtering, control, sensors, simulation, RL, IL, and safety. There were 1-2 coding interviews, pretty standard Python questions like syncing to a mock database. No tough LeetCode stuff, just needed to be fast, think it through, and explain my approach. One ML interview/quiz covered ML basics such as loss functions, classification vs. regression, dimensionality reduction, regularization, CNNs, RL, and model training. Then there was a systems design interview, a brainstorm session about designing a mobile manipulator teleoperation system, with some deeper questions on specific areas and trade-offs.

Applied online, and about 3 weeks later, I got an email from a recruiter asking for my availability for a call, and they also mentioned an NDA. I sent over my availability for two weeks starting the following Monday. However, the recruiter never scheduled the call or sent the NDA. I followed up on Thursday, and the recruiter said she was traveling the next week and asked if I could do the call that Friday or if we could find another time. I confirmed I could do it on Friday, but again, I received no response. I waited two more weeks thinking she might be on vacation, but still heard nothing. I sent another follow-up email, still no reply. I then reapplied online and got an automated rejection email stating a cooldown period, which was confusing since I never actually got notified of a rejection, only that the call wasn't scheduled. It felt like they were just collecting data, which was really disappointing from a company like Google.