We set out to build a different kind of technology company in 2015. A company that was building for the present and the future at the same time. We wanted to innovate on our organizational design and communication apart from technology innovation. Also, we needed new ways to hire and retain talent here. The existing methods didn’t work quite well for us in our initial months. We have gone to great lengths to hire outliers and misfits that fit our culture and believe in our vision. This post is about how we want to crack ML hiring here.
At Artifacia, we have built a ground-breaking visual intelligence platform currently focused on retailers, and is supported by cutting-edge work going on at Artifacia Research, our lab where we work on lots of interesting and hard problems in AI and Human Computer Interaction. It’s headquartered in Bangalore, India. With a globally recognized team of research and software engineers well-versed in the nuances of deep learning, reinforcement learning and GPU programming, we have established ourselves as the best AI startup in the country and one of the best in the world in a very short time. We can say this based on our performance in popular AI competitions like MSCOCO, Kaggle competitions, recognition from companies such as Google, IBM and NVIDIA, and feedback received by our innovative product offering.
We probably get more applications for ML everyday than most other companies in the country. That includes applications from candidates with previous experience at companies like Google, Microsoft and Xerox and top universities from the US and Europe. This might be due to some of our well-known work including our flagship research project named Project Turing and my public appearance in various events around AI and entrepreneurship. And probably it has a lot more to do with an all-time-high hype around AI. Hype leads to misinformation and many people stop asking themselves hard questions before jumping into a new field. This happened with Big Data/Data Science and now it’s happening with AI. In my experience, just like Data Science we have one great talent against several mediocre in AI. The very best talent I’ve come across from across the world in such niche fields are mostly from very good labs from top research universities or big tech companies investing heavily in AI research. Then there is a relatively small percentage of talent pool which is primarily self-taught using publicly available courses, tools and knowledge online.
Our stats over the last year reveal that the success rate for our research focused hiring is currently less than 1%. Not something we are proud of but it shows our focus on only the very best talent. It’s definitely more competitive than getting into an IIT in the country or getting into a Google or Facebook. This has a lot to do with questions asked by Vivek - who heads research here - and myself in our fairly elaborate selection process for research positions here. We have found a lot of issues that hinder the chances of success in our rounds and we thought it would be great if we could come up with guidelines on how a good candidate can go about cracking our rounds here.
Step 1 : Write to us with a simple reason behind why you want to work with us. Keep it small and avoid jargon.
Step 2 : If you are applying for a Research Engineer position, you should ideally have an MS in AI/ML/CV. This can also be relaxed to BE level if you are out-of-this-world crazy good.
Step 3 : If you are applying for a Research Scientist position, you should ideally have a PhD in AI/ML/CV. This can also be relaxed to MS level if you are out-of-this-world crazy good.
Step 4 : Send links to your projects in ML or papers that you think are fairly high-quality. A good profile on Kaggle will help too.
Step 5 : If your application looks impressive to the founders here, you will get an introductory call from either Vivek or Navneet.
Step 6 : Based on the call, you maybe asked for a Kaggle round to begin with. Don’t use Matlab. It’s not a proper programming language.
Step 7 : If you do well, your second round will be an intensive technical round based on what you did in the first round.
Step 8 : The next step is a two-part online algorithmic round. Programming skills and mathematics are crucial to being good in ML.
Step 9 : Based on your performance, you will be taken to the next round which is again a programming round on Skype/phone/whatever.
Step 10 : You may be asked for the final on-site interview. Do well in that and get the most prestigious ML job in the country.
I hope this guide will be helpful for people who want to work at Artifacia Research. I’ll keep on updating it as we learn new ways to improve our research hiring.
P.S. 1. I’ve used ML in several places instead of AI/ML/CV for the sake of simplicity. Modern AI (Deep Learning) is anyway mostly statistical ML which is impacting other fields like CV, NLP and Robotics.
P.P.S. This can also be used by other startups for great hiring in ML. In our experience, one great ML person is better than 10 good ones.