Former intern turned trainee manager Richard Everett describes his journey to DeepMind, sharing tips and advice for aspiring DeepMinders. Applications for the 2023 Internship will open on 16th September, please visit https://dpmd.ai/internshipsatdeepmind for more information.
What is your way to DeepMind?
Like many people, I loved playing multiplayer video games growing up. I was fascinated by the interactions between human players and seemingly intelligent computer-controlled players, and I dreamed of a career in AI. This dream led me to pursue an undergraduate degree in computer science. A common (but not exclusive!) avenue in industry. However, after working on several research projects with my professors, I developed a taste for research and decided to go on to get my Ph.D.
Around the time I started my PhD, a small startup called DeepMind was acquired by Google. As I looked more closely at their research, I quickly found that it inspired my own research, and so in 2016 I decided to apply for an internship. After a few interviews with engineers, researchers, and program managers, I didn’t get an offer. However, after meeting a group of great researchers, I decided to reapply the following year and took the internship. That experience led to a full-time offer and I’ve been here ever since, working on AI and helping interns who are going through the same experience.
Can you describe the training interview process?
The interview process was exhaustive, but it has evolved since I applied. Today’s interns can expect the entire process to last only a few months, which includes a technical interview and a group interview. In my application, I listed researchers with whom I was particularly interested in working, and with whom I was fortunate enough to speak after my technical interview. I was so excited. This was a unique opportunity to talk about my previous work and brainstorm potential training projects with world-class researchers I’ve followed for years, asking them questions about DeepMind.
Recruiters have been incredibly helpful in walking me through the process and providing resources to help prepare for interviews. For the technical interview, I prepared by reviewing my first-year undergraduate courses in mathematics, statistics, and computer science. For example, reviewing linear algebra, calculus, probability, algorithms, and data structures. I also did some coding exercises where I tried to talk about what I was doing.
For team interviews, I reviewed the team’s recent work (ie papers, blog posts, articles, talks), and thought about how my work might relate to it. I also came up with a short list of questions I wanted to know more about, such as the team’s collaborative style and how previous training sessions worked.
What was it like when you joined full time?
It took me a long time to find my footing! With so many exciting projects underway and great people to talk to, working at DeepMind often feels like a kid in the world’s largest candy store. For interns, developing and focusing on just one project among many is challenging, especially within a limited time. This was a challenge I found in my internship, and today I enjoy supporting new beginners through the process who are experiencing the same excitement as their first time.
Why did you get involved in the training program as a full-time employee?
Having gone through the coaching experience myself, I can relate to what our current and aspiring trainees go through. It can be nerve-wracking, exciting, confusing, and inspiring, all at the same time. After receiving a lot of support during the training period, I wanted to provide the same support to future trainees. As a result, I am now coordinating the internship program for my team and I am in several groups constantly striving to improve the program via DeepMind. I also interview, mentor, and manage interns, as well as spend time networking and speaking with potential candidates (eg at GraceHopper, NeurIPS, and research talks).
What kind of work do the trainees do?
It is always interesting to hear what interns decide to pursue during their time with us. In my team (Game Theory and Multi-Agent), we work closely with our interns to co-develop projects that they can make themselves, and this has resulted in an incredible array of projects over the years.
To highlight just a few general examples, trainees designed new multi-agent environments (eg inspired by the social deduction game Between Us and Assembly Lines), developed infrastructure to study human factor interaction, used cooperative game theory for language and team negotiation models, and worked on reverse reinforcement learning. Multifactorial, revealing antagonistic examples of reinforcement learning, mastering the game strategy, and applying evolutionary game theory to online learning.
How would you describe the culture at DeepMind? and your team?
In short – kind and cooperative. Over the years, I’ve heard dozens of interns and new starters make the same remark: “I can’t believe how friendly and supportive everyone is!”. The amount of time, energy, and support DeepMinders give each other is amazing, and this extends all the way from veterans of the company to new starters on day one. Everyone is always happy to grab a coffee to chat, discuss their work, share feedback and partner together on projects.
For example, one of my favorite projects at DeepMind (Powerful Cultural Transfer Learning in Real-Time Without Human Data), came from a close collaboration between artists, designers, ethicists, program managers, QA testers, scientists, software engineers, and research engineers, over two years. This diverse and collaborative culture also extends to our internships, as internship projects typically involve many collaborators and advisors from across the company (involving roles, teams, and even offices!). For example, many of our Game Theory and Multi-Agent interns work closely with DeepMinders from the London and Paris offices.
From left to right, a subset of the project’s authors: Ashley Edwards (RS, London), Merona Beslar (RE, Paris), Corey Mathewson (RS, Montreal), Alexander Zakirell (Designer, London), Richard Everett (RS, London) ), Edward Hughes (London, London), Avishkar Bopchand (London, London).
Any tips for aspiring DeepMind interns?
For students interested in AI, there are plenty of easily accessible resources for independently learning more about the industry and DeepMind: from papers, blog posts, and talks to open source code, demos, and tutorials. It’s easier than ever to get stuck! You can also participate in workshops and conferences, many of which offer student discounts and mentorship opportunities (such as Deep Learning Indaba, Cooperative AI). For me, I found my love for AI research by talking to professors about their research between classes, working on projects with them, and then networking with other researchers in areas I’m excited about.
DeepMind is made up of kind, helpful, and driven people from all walks of life, and our training program reflects that. Whether you are an undergraduate or PhD student, studying a technical, physical or social science subject, with or without AI/ML experience, there is likely to be an internship opportunity for you. We offer courses across different teams in Research, Engineering, Science, Ethics, Society and Operations.
Having gone through the process myself (twice even), I can completely understand and relate to how intimidating the app is. I’ve spoken to many incredibly talented students who wrongly believe DeepMind is out of reach or that their skills are insufficient, and therefore not applicable to them. If you are thinking of applying for an internship, my real advice to you is to just do so. You have nothing to lose, and you and DeepMind probably have a lot to gain.