in

From LEGO competitions to DeepMind’s robotics lab


Right this moment’s submit is all about Akhil Raju, a software program engineer on the robotics workforce. We initially met Akhil in season two of DeepMind: The Podcast, however we needed to get to know him higher and listen to extra about his path to DeepMind.

What sparked your curiosity in synthetic intelligence (AI)?

After I was younger, I thought of AI in the identical manner I thought of magic. Sure, I needed to hang around with R2-D2 and Optimus Prime – however I additionally needed to go to Hogwarts. That was till I turned 12 and began taking part in LEGO robotics competitions. At that second, I discovered that robots weren’t a fantasy or one thing that might solely exist within the far future, however fairly one thing that may very well be created within the current. Additionally, it seems that taking part in with robots is extremely enjoyable.

What got here subsequent?

Loads! From there I continued with robotics competitions, began college at MIT, and spent plenty of time learning pc science with a concentrate on robotics. After commencement, I fully pivoted away from the sphere and joined a startup in San Francisco for a number of years earlier than transferring to Google.

It was nice however I would all the time needed to stay overseas so I began taking a look at alternatives outdoors the US. At that time, I made a decision to maneuver to London and set my sights on DeepMind. I really didn’t suppose DeepMind employed folks with out PhDs, however I gave it a shot and it labored out!

As a result of I used to be doing a switch from Google to DeepMind, I used to be capable of apply to a number of groups on the identical time. The robotics workforce wasn’t on my radar till my recruiter requested me, “By the way in which, you might have bits of robotics in your resume. Have you considered becoming a member of our robotics workforce?” I bit on the alternative. And actually, it’s been superb ever since.

Are you able to stroll us by means of a typical day? What’s your schedule like?

Each morning I head into the workplace for breakfast, the place with out fail, my teammates already are. It’s grow to be part of our day by day routine to have breakfast collectively earlier than leaping into our work.

I spend most mornings within the robotics lab, fixing failures from earlier experiments or establishing new robots. Even when there’s not a lot to be executed, I get power from simply strolling round and seeing our robots at work, listening to the hum of the machines and motors. We’ve grown quite a bit over the previous few years, and you may really feel it whereas strolling round our house.

My afternoons are a mixture of conferences, coding and – now that most individuals are again within the workplace – an impromptu chat or two. That’s one in all my favorite elements of being within the workplace – the random catch-ups and whiteboard periods that assist me study and transfer rapidly. From there I’ll take a fast snack break, and if the climate is sweet, head to the balcony to compensate for a few of my favorite US sports activities podcasts (I nonetheless haven’t made the swap from soccer to soccer). Then I’ll code a short time longer.

What’s it like behind the scenes at DeepMind? 

The tradition at DeepMind is likely one of the greatest elements of being right here. From my perspective, we’ve discovered a pleasant stability between a college, start-up, and enormous firm. A lot of the work tradition comes from the primary two.

It’s commonplace to seek out folks brainstorming in entrance of whiteboards with maths scrawled throughout it, or somebody tucked away in a quiet nook studying the most recent analysis papers. Much like a start-up, there’s a palpable power all through – you possibly can actually really feel everybody’s pleasure.

It might be cliché, however whenever you love what you do it would not really feel like work. The robotics workforce is a miniature model of all of this, with the bonus that many people are shut pals outdoors of labor, too. It’s good.

How was working from house through the pandemic? 

Like most individuals, I spent the primary month of the pandemic in disbelief, assuming we’d be again to regular quickly. Nearly all of our conferences and collaborations moved on-line which was an attention-grabbing expertise for our workforce specifically.

As soon as the realisation set in that we have been on this for the lengthy haul, I made a decision to spend my new-found free time to higher myself. I attempted a bunch of hobbies – lengthy sufficient in every case to say I’d tried it, however not lengthy sufficient for something to stay. I had my guitar section, a cooking section, and even a puzzle section, however my favorite was the tie-dye section. There have been a number of weeks the place I tie-dyed all the things, from shirts to shorts to socks, and now they sit on the backside of my closet (the place they actually belong).

What sort of optimistic impression are you hoping AI can have on the world?

I really feel fortunate to be at DeepMind and to have the ability to concentrate on the work that I do. Robotics – and AI usually – will probably be a optimistic pressure on the earth, and it’s thrilling to have the ability to assist transfer that ahead.

On the entire, I’m notably focused on seeing how AI can assist mitigate local weather change – whether or not that’s by discovering methods to make use of power extra effectively, or enabling us to provide clear power. Researchers at DeepMind are already fascinated with this, so I’m hopeful that we’ll be capable of transfer the world ahead and make an impression on this house.

Any ideas for aspiring DeepMinders? Or somebody seeking to get into an identical position?

If you wish to be at DeepMind, go for it. Apply, interview, and simply strive. You won’t get it the primary time however that doesn’t imply you possibly can’t strive once more. I didn’t suppose I might get a job at DeepMind, and after I obtained the supply, my preliminary thought was – certainly this can be a mistake! Everybody doubts themselves – I’ve by no means felt like the neatest individual within the room. I’ve typically felt the alternative. However I’ve discovered that, regardless of these emotions, I do belong and I do need to work at a spot like this. And that journey, for me, began with simply making an attempt.

Be taught extra about robotics at DeepMind and seek for open roles at present


Open-sourcing MuJoCo

Emergent Bartering Behaviour in Multi-Agent Reinforcement Studying