How to interview a Data Scientist

Looking at hiring a Data Scientist in 2015? Here are some tips on interviewing


You know the feeling. You’re sat across the table from an interviewee and the conversation starts to run away from you. Your eyes start scanning the room as they’re reeling off another jargon-crammed answer. You notice the chip in the Channing Tatum mug you received from the office Secret Santa last Christmas. You then notice the scratch on your hand that your 4 year old gave you, in exchange for closing the laptop during that far too familiar One Direction song. The only thing you’re not noticing are the answers the other person in the room has been reeling off at you.

If this typically happens to you most when you’re conducting technical interviews, it could be either that the person your interviewing is talking in a language not known to you, or you simply don’t have the right questions to ask that cut through the techno-babble and get to the point of why they're sat there in front of you.

The polar ice caps of the ‘Data Scientist Age’ have well and truly melted, so if you are one of those countless companies who are currently looking for one, there’s a chance you’re going to have one of these interviews.

The subject of this article probably won’t offer any value to those of you reading who know your way around a Random Forest, or a Neural Network, but it’ll hopefully give a few pointers to those of you who don’t.

As I mentioned earlier, the age of the Data scientist is well and truly here. This means that are highly talented geniuses in our population who can change the landscape of an entire organisation, through the development of an algorithm and the implementation of some code. This also unfortunately means that there are lots of people out there who want you to think that they are one of those geniuses too. The question is, can you tell the difference?

Having spent the last year interviewing a large number of Data Scientists, I’ve developed a simple set of questions that help me to understand the what, the why and the how of what they do.

The What

At the very start of the interview, I normally like to get a general feel for who I’m speaking with. “What projects are you most proud of?” “What contributions have you made to the businesses you’ve worked for?” “Can you describe in detail the responsibilities you’re looking for in your next position?”

When interviewing someone involved in post-doc/PhD projects, I’ll always look to get an understanding of the projects they’ve encountered there, however, it’s really important to understand the time constraints they’ve been under, as typically, someone solving problems in academia may be given far more time than you would encounter in the commercial world.

The Why

The best Data Scientists I’ve worked with have all had substantive expertise in a particular domain, or at the very least, will understand what the business impact is of the work they have doing.

How can you solve the problems the business is facing, if you firstly don’t understand the business itself? The same sometimes goes for domain too. However, it’s very important to note that a good Data Scientist will always be good at problem solving, no matter what domain they’re working in, so don’t get too hung up if the person you’re interviewing doesn’t work in your exact industry.

I’ll typically ask questions like, “What were the business outcomes of the projects you worked on?” “Give me an example of when you’ve thought about the businesses product?” “Tell me about a time when you’ve improved a business process?”

The How

This is the part of the conversation where I’ll begin to understand whether or not the person I’m speaking with, is the person my client is looking for.

It’s very important to note that there are lots of people with the title ‘Data Scientist’ who can’t write Machine Learning algorithms, can’t code…or can’t do either.

What you may find, is some Data Scientists who use Machine Learning libraries that have been written by other people in the team, filled with algorithms they use like tracing paper. “Can you give me an example of when you’ve written a unique algorithm?” “Can you give me an example of when you’ve developed an algorithm from a framework/research paper?” “Can you give me an example of when you’ve written/implemented your own code?”, these are questions to help you understand if you’re speaking to someone who can create things from scratch, as these are typically the ‘A Players’ you’ll want to hire.

It’s worth pointing out that if you find someone who has nailed all of the above questions and you have that gut feel that they may do wonders for your business, please don’t get too precious about culture, team fit, etc. Don't get me wrong, these things are important, but people like this can be incredibly hard to find…sometimes harder than finding that missing piece of Mr Tatum.


comments powered byDisqus

Read next:

Groupon: What Advice Would You Give To Someone Starting A Career In Big Data?