This post is part of a series in which I address common questions I get from data-interested executives and data practitioners.
People often ask me how to prepare for a job interview in the data science/machine learning/AI space. Since these are technical roles, it’s tempting to focus on the technical interview: “what should I study?” “should I know about [some technology]?” “how well do I need to know [some algorithm]?”
Your homework starts well in advance of the formal, on-site interview … and the technical portion plays just a small part.
I’ll walk you through six questions to consider as you progress from scanning a job posting to participating in the on-site interview. Providing honest answers will save you time and effort by narrowing your focus to the best opportunities.
These questions are useful even (or, perhaps, “especially”) if the company reached out to you instead of the other way around.
You’ll be most effective as a data scientist if you have knowledge of, and interest in, what the company does.
Say you’re a sports fan. Wouldn’t you rather be a data scientist for a major sports team? Maybe your parents work in real estate, and you’ve learned a lot through your passive exposure to their field.
Having such a connection will put you ahead of those who don’t: you’ll have a hunch as to what are meaningful results before presenting them to stakeholders. You’ll be able to anticipate business units’ needs. You will be poised to spot new opportunities for the company to collect, monetize, and analyze data. Over time, you will be less of “just a data scientist” in their eyes and more of a “domain expert who just happens to know how to use and analyze data to solve our problems.”
Companies that plan their data efforts – everything from a well-informed idea on how to apply machine learning to their business problems, all the way to a formal, company-wide data strategy and road map – before hiring a data scientist are serious about using their data to drive business improvements. They are trying to position themselves (and, therefore, you) for success.
The job description is your first indicator of whether a company has developed a plan. Look for specific uses of data science, and compare that to the technical skills they seek. A vague job description, followed by a list of machine learning buzzwords, is usually a sign of a company that is confused.
Sometimes you’ll have to speak to someone in-person to learn more about the plan. I’ll explain that in the next section.
Terms like “data scientist” and “machine learning engineer” don’t have clear definitions. You’d do well to learn the specifics of the role in question, such as:
- Is machine learning a core component of the business model, or in a support role for other business functions?
- What are the role’s day-to-day activities and responsibilities?
- What does the current data team look like? and how is it expected to grow and mature over time?
- Is this job the first data hire? Are you expected to build out the rest of the data team?
These can be tough to sort out from just the job posting. Speaking to someone inside the company – ideally, the actual hiring manager – would be the best way to get this information. You can start by exploring your professional network to find a connection to such a person. In some cases, you won’t meet them until you get an on-site interview.
Pay scales vary by location, company, and role. Salary surveys and salary comparison websites will provide rough ideas of numbers, which can set the tone as you negotiate terms.
I defer to Patrick McKenzie’s in-depth blog post on this topic for more details. Everything I could say here, he’s already said in much greater detail.
Technical interview practices vary by company, from grueling whiteboard interviews, to puzzler questions, to a quick chat over drinks. Some throw in a take-home project for good measure.
It would be difficult to prepare for every possible approach. You’re better off figuring out what’s right for you, and using that as a filter: when the company reaches out to scehdule time with you, ask what to expect in the interview and with whom you’ll be meeting. You can then prepare accordingly or, if need be, decline to pursue the role.
The terms “culture” and “fit” have earned some well-deserved flak, because they’re often euphemisms for unfair reasons to reject a candidate.
Taken at their textbook definitions, though, these are important concepts: you’re going to be in this company and with this group for (at least) 40 hours per week. You want a job where you can feel like part of the team, where you’ll be productive and comfortable, and where your career can grow.
Do yourself a favor and use the in-person interview to assess this company for fit and culture:
- Do the people meeting with you seem organized and welcoming?
- Do you feel comfortable around the team? Does their style of work match yours?
- Do the people around you look busy-but-relaxed? goofing off? stressed?
- Do you get the impression that everyone on the team is getting along?
Similarly, don’t be afraid to simply walk out of an interview gone awry. If someone is being rude or otherwise unpleasant to you now, during the interview … there’s little chance they’ll shape up after you join the team.
It sometimes feels as though every company wants to hire more data scientists, which means there are plenty of jobs from which to choose. Be sure to do your homework and find the role that is best for you.