Data Science Interview Questions: Ace your interview

After weeks of applying for data science jobs, you get that email you’ve been waiting for: A recruiter has invited you to interview for the role! And if you can get through the interview process for what the Harvard Business Review called the sexiest job of the 21st century, you won’t only change your career. You might change your life, too.

And with GlassDoor ranking data science as #3 on its most recent list of Best Jobs in America, the demand for data scientists is only expected to rise. [What is data science? Read more here.]

For more on what to expect from the interview process—and how to ace answering common data scientist interview questions—we’ve consulted data scientist Eric Feder. In a career spanning over a decade, he’s managed large data science teams and once did statistical analysis for the New York Yankees. He has interviewed 250 data scientists for entry-level to senior managerial roles, so it’s an understatement to say he really knows data science interview questions.

Let’s get to it.

How can you crush your data science interview?

Mine the data in the job description

Data science generally refers to the ability to use statistics, algorithms, and other tools to extract insights from data and interpret them for a wide audience. But within this wide-ranging, rapidly-changing field, the same terminology can describe many different kinds of roles. In the way that “cook” can refer to a local sushi chef, a Food Network host, or the person at your high school cafeteria, “data scientist” can mean very different things in different contexts.

Feder points out that for all positions, “The best way to prep is to read the job description and talk to the recruiter.” Look at the information they’ve already shared with you and mine all available resources for insights, just like you would with a data set. 

The first place to look? The job title. Here are some of the most common ones, with commentary from Feder:

• Data Analyst: Gathers, cleans, and preps data and then gleans business insights
• Data Scientist: Similar to the above, but with more tools, advanced mathematical models, and sometimes multiple data sources
• Data Engineer: Less about statistics and insights, and more about managing data sets and building pipelines: “At a high level, if you see a data engineer posting, that’s an engineering-heavy role: You won’t need stats or math as much, and it’s building out a pipeline or managing data. Those questions will be more about building out pipelines or managing large data sets.”
• Specialized roles that focus on machine learning (ML), artificial intelligence (AI), or natural language processing (NLP) usually list those in the job title. “A lot of companies now are using the ML engineer to refer to what a few years ago would have been called a data scientist: They’re building and testing models and deploying software to test those models,” Feder says. “It’s getting popular as a way to differentiate between statistics and machine learning.”

Next, read the job description itself. What skills are absolutely required? What are nice-to-have? In hiring for this specific role, what business needs might the team be trying to address? Is the team hiring urgently? If there’s anything that would prevent you from taking the job, such as location, clarify exact requirements with the recruiter. 

What’s included in a data science interview?

Though roles vary, most data science interviews have similar components. The process usually includes a recruiter call, technical interview (which can include an asynchronous technical assessment, a live interview, or both), and a behavioral interview, though not always in that order. 

Let’s look at data scientist interview stages. 

Data Science Technical Interview 

During technical interviews, companies are looking for two skills: technical skill and the ability to communicate insights. That last part is what makes the data science interview process different than other engineering roles.

“Baseline technical skills are a must-have,” explains Feder. “That can cover statistics, machine learning, and math or other engineering areas. Every company is going to be testing your programming or SQL, and sometimes both.” 

Common technical data science interview questions include:

• “What is the difference between univariate, bivariate, and multivariate analysis?”
• “How do you deal with missing values in a data set’s variables?”
• “What are the steps to maintain a deployed model?”
• “What algorithms are used for recommender models, for example to shoppers on Amazon?”

Presentation is also paramount. “They’re looking for basic communication and storytelling. They want to know that you can explain yourself and talk through basic technical concepts.” This demonstrates that you’ll be able to carry over your communication skills to the workplace, where you’ll need to share the insights your data analysis reveals.

One strategy for addressing both the technical and storytelling sides, he says, is to describe why you made certain decisions. “If I’m interviewing someone who is going to do ML for their job, I’ll always ask, ‘How do you know your model is working?’ There are technical approaches that you learn in school, but newer data scientists should be able to describe how to apply them.” 

Show your work, too. “Having concrete work output to point to—GitHub, portfolio—shows that you can take a project from start to finish and incorporate it in a report,” Feder says. “It also lets hiring managers check the codebase, model, and data analysis.”

Finally, you may need to be ready to nail the communication and presentation piece ready sooner rather than later. Feder explains, “Companies seem to be using the technical interview as a screening. A few years ago, they would have started with more of the behavioral interview or stats.”

So while you may think of data science roles as being all tech, don’t discount professional skills. 

(Did you know that BloomTech courses, including the Data Science program, teach professional skills as well as ML, SQL, and other in-demand skills?)

Behavioral Interviews for Data Science 

Particularly for data science, the non-technical interview shouldn’t be overlooked. Eric advises paying particular attention to generic behavioral questions. Expect some of these common behavioral data science interview questions:

• “Tell me about a time when you used data to influence a business decision.”
• “Talk me through a time when your approach to a problem didn’t work.”
• “Tell me about a time when you disagreed with how a customer or stakeholder applied insights from your analysis. What did you do?” 
• “Walk me through a project you worked on as part of a team or cross-functional effort. How was the project organize, how were tasks delegated, what role did you play, and how did you coordinate with the rest of the team?”
• “Tell me about a time when you needed to communicate results with a non-technical audience.”

“Preparing for [data science interview questions] makes a difference and really matters for a data science job,” Feder says.

Behavioral questions are a terrific time to showcase previous hands-on experience. “They’ll want to hear you talk about a project that you’ve done, whether that’s in an industry or academic setting,” says Feder. “What was the project? Why did you work on it? What was interesting? What was challenging? What were your findings? They want to know that you can apply your theoretical knowledge to a practical question.” 

The thing to remember about behavioral interview questions is that they’re not all about your hard skills. This portion of the data science interview is meant to tease out how you’d fit into the team or company, too.

“I really like any kind of internship experience, whether that’s in data science or any other experience that shows that you understand how to be a part of an organization, even if the work itself isn’t the most applicable to the role that you’re applying for,” Feder says. It’s helpful if that previous work was at a similar organization or industry, but not entirely necessary.

And don’t forget the bigger picture, he says. He has seen otherwise well-qualified candidates be eliminated for not being able to answer why they’re interested in the company, the role, and why they’d be a good fit for both. He continues, “Companies genuinely want you, as a candidate, to be interested in the company. There’s a lot of dimensions that companies are evaluating your fit for, so: Why do you want to work here?”

(Still nervous about your performance during interviews? BloomTech’s Data Science course includes built-in career support starting on your first day of class. Learners can get interview prep before every interview they land, and career coaches even help identify jobs that are a great fit for learners!)

Avoid This Interview Mistake

No matter what role you’re interviewing for, one thing holds true: “The worst thing you can do,” says Feder, “is state on a resume that you have a skill or set of knowledge that you don’t have. Anything on a resume is fair game for an interview to ask about. If you come out of an interview and can’t discuss something listed on a resume, that reflects poorly on you.”

Then again, that’s true outside of data science, too. Honestly and transparency is the best policy for your career. 

If you need to brush up on your data science skills—or even learn them from scratch—consider enrolling in BloomTech’s Data Science course. Our expert instructors teach Python, SQL, linear algebra, and statistical analysis, data visualization, deep learning and ML ops, and much more.

BloomTech’s courses are built upon hands-on projects that not only teach you the technical skills, but also how to apply them to real problems. What’s more, you’ll work on a real project for a real client, where you’ll hone professional skills like communication, project management, stakeholder reviews, and more. Remember, these are precisely the skills that could set you apart when interviewing for a data science or data analytics role. 

Curious? Test out BloomTech with no financial commitment during our Risk-Free Trial. You’ll get 3 weeks to experience how we approach training the next generation of data scientists. (Spoiler alert: We’re committed to teaching the skills most in-demand by employers through industry-leading curriculum, live academic support, and career coaching that begins as soon as you enroll.) Sign up for your Risk-Free Trial here!