Are you overwhelmed by data and desperate for information you can use? Have you decided to hire a data scientist? Please don’t drive away the best candidates with inappropriate data scientist interview questions and behavior.
As a hiring manager, you’re probably aware of competition from other employers who want to hire a data scientist. If you’ve been concentrating on data scientist salaries, benefits and career paths, you know it can be tough to compete in those areas.
What you may not know is that many of those competing employers are doing a great job of driving away the best candidates for data science jobs. What you need is ways to set yourself apart from those competing employers.
You can start with the same positive approaches you would use to fill any other role. Simple basics can help you attract more candidates for your data science job:
- Network for candidates: People in your network know and respect you. They’ll present you to prospective hires in a positive way.
- Interview at schools your competitors ignore: There are more than 5000 colleges in the United States. Perhaps you have missed some good sources for new talent.
- Think beyond common keywords: While the data scientist title has become popular, there’s no official definition for it, and it may not even be the best choice for your situation. Explore related terms like “data analyst,” “statistician” and others to see what works best in your ads and data scientist job description.
When it comes time to interview, you need to make yourself and your data scientist job appeal to the candidate. In concept, this is pretty easy - just don’t be a jerk. In practice, a lot of hiring managers and recruiters seem to find that difficult.
Every candidate who’s interviewed for data scientist jobs has stories of backing out at the interview stage. Those stories get around. If the details of your interviews are going to end up on Glassdoor, plan ahead and make sure they’ll be good for your image.
Avoid these 5 common “Don’ts” for data scientist job interviews:
1: Don’t demand that candidates know what you know
One of the most common ways to drive away candidates is to bombard them with very specific technical questions. That barrage of questions has more to do with interviewer wanting to feel smart than learning what the candidate knows.
It’s better to ask candidates to explain more about areas of expertise mentioned on the resume. If you need expertise that’s not mentioned, open with broad questions and give the candidate a chance to introduce detail as appropriate.
2: Don’t put technology before business
Want to make a good data scientist shudder with just four words? Hand the candidate a whiteboard and say, “Let’s see some code.” (That’s a direct quote from a real interviewer.) Nobody does great work while wearing interview clothes and being watched.
Code, tools and algorithms are just means to provide information that solves business problems. Don’t make them out to be more than that.
Consider technology in context. Ask candidates about business problems they’ve solved, technology they used along the way, whether they’d do it the same or a different way next time. Probe for willingness to learn and use new methods in the future as need.
It may be appropriate to ask to see samples of code and other past work, but asking them to perform on the spur of the moment is just mean.
3: Don’t insult candidates
An experienced and very good analytics recruiter shared his worst-ever client story with me. The client seated the candidate, a heavy man, in a flimsy chair, which collapsed. Despite the obvious embarrassment and potential to injure the candidate, the client did this a second time later in the day.
Although I’ve never heard another story quite that bad, I have heard plenty of stories of interviewers remarking on a candidate’s weight, fashion sense and other personal matters unrelated to data analysis competency. But you know better than to do something like that, right?
Most insults are more subtle, and interviewers don’t always realize that they’re doing something offensive. Data scientists are looking out for these warning signals that hint at culture problems:
- Don’t say things that aren’t strictly true, especially not the ever-popular, “The starting salary may be low, but there are opportunities for growth.”
- Too personal: We must have your salary history. Where else are you interviewing?
- Borderline illegal: “Do you think Jesus would be a Mac, Windows, or a Linux user?” Are you married? Do you have kids?
- Brain teasers: Puzzles, riddles, anything resembling a math homework problem.
- Silly: Google “data scientist interview questions” and you’ll find dozens of great ways to waste time in an interview, such as, “If you were a tree (can of soup, fruit, vegetable, house, etc.) what kind would you be?”
- Biased: Avoid anything geared to detect “fake” data scientists – these are nothing more than ways to reinforce one manager’s world view. Also, do some reading on gender, race and religious bias and explore ways to avoid unintended bias in your hiring process.
- Rude: Don’t say nasty things about the candidate, other candidates, departing staff or anybody else. Please.
4: Don’t demand free work
It’s appropriate to ask a candidate about past work or to review some of that work if sharing won’t intrude on the privacy of another employer or client. It’s not appropriate to ask a candidate to do new work unpaid.
That means you have no business asking a data scientist job candidate to complete a project assignment, download and demo your product, or provide details of how to solve your real-life problem.
5: Don’t be boring
Perhaps the worst possible thing you can do when hiring data scientists is to be boring.
The best way to be boring is to focus too much on yourself. A good data analyst wants to know about you and what you do, but there’s such a thing as overdoing it.
If you leap into long product demos right away, you’re boring. If you wax lyrical about yourself, you’re boring. If you get way off topic for a long time, you’re boring.
It’s not as hard to hire a data scientist as you may think. If you offer a living wage, decent benefits and interesting work, there’s good analytics talent for you. Just make sure you don’t get in your own way.
