person arriving for an interview and shaking hands
shapecharge/Getty Images

With the massive push toward using meaningful data—not to mention the tremendous amount of data produced thanks to new technologies—it only makes sense that a new segment of workers would be needed to manage and analyze all of that information.

That’s where data science comes in. The field is still rather new and constantly evolving. But at a core level, data science seeks to analyze available data and use those analyses to develop useful conclusions.

So how do you prove to a prospective employer you’re a worthy data science hire who can help their organization succeed by making sense of all the data they’re being inundated with every single day?

If you’re getting ready for a data science interview, there are a few key things you need to know and several questions you should be prepared to answer.


Pro Tip #1: Understand Which Kind of Data Science Role You’re Interviewing For

Some experts predict that the title “data scientist” will be phased out by the end of the decade; however, data science will remain relevant as a core business function—your title just may be something like “product analyst” or “data engineer.”

“Data scientist refers to a spectrum of positions depending on industry, organization, and role within an organization,” says Shanshan Ding, Director of Data Science at Hinge. “When you apply to be a data scientist, you could be applying to be an NLP researcher, econometrician, product analyst, machine learning engineer, data engineer, or really any position that generates, transforms, analyzes, and/or productizes data.” (Full disclosure: Hinge and Wayfair, below, are current clients of The Muse.)

Because of the wide range of jobs included in data science, it’s important to really pay attention to the employers you’re interested in and the job postings you’re responding to. While one job might be super statistics-heavy, another might be much more focused on programming or a different component of data science. Analyzing this information will help you prepare for interviews and know how to tailor your answers to that specific role and company.


Pro Tip #2: View All Questions Through a Technical Lens

Candidates are often passed over because they don’t seem “technical enough,” says Max Dosad, Strategic Accounts Manager at Harnham, a recruiting company that focuses on data and artificial intelligence roles. Yet at the same time, those very same candidates might feel that they weren’t asked many technical questions.

Dosad’s advice? No matter how non-technical a question seems, try to view it from a technical perspective. Think about it this way: How can you use each question to demonstrate your technical skills and abilities?

Let’s say a hiring manager asks you a question about how you would handle a certain business problem. While it may not seem, on its face, to be a technical question, they’re still looking to see how you would analyze the problem, create a technical solution to resolve it, and then communicate that solution to non-technical stakeholders.

Incorporating technical elements into your answers throughout an interview will ensure that you don’t miss out on any opportunity to showcase your technical acumen and analytical abilities.


Pro Tip #3: Be Ready for Open-Ended Questions

Hiring managers and recruiters won’t necessarily ask many “academic” or “textbook” questions. Instead, they’ll likely present you with broad, real-world, open-ended questions, business problems, or case studies.

For example, Evan Butters, a data science recruiter at Wayfair, asks questions that are related to a challenge that’s actually being worked on at the company and then assesses how the candidates would go about addressing it. This opens up a conversation and allows managers to see exactly how you’d work as part of the actual team.

Open-ended questions allow candidates to demonstrate their problem-solving skills and require them to:

  • Understand the problem. What exactly is the issue and what is the end goal that you’re seeking to achieve? Why is this problem important to solve for this particular business and industry?
  • Define the boundaries. In your answer, outline the scope of the problem and any assumptions you’ve undertaken in order to address it. For example, is the solution generalizable, or would it only work in this specific instance? Why?
  • Discuss the trade-offs. Your ability to think through and articulate the pros and cons of different possible solutions is the most important part of your answer, Butters says, so weigh this heavily in your response.

There are no “right” answers here, and there’s likely no “best” answer, either. There are multiple ways to go about solving problems and what interviewers are really looking for is how you’d approach them. Butters also reassures candidates that Wayfair isn’t necessarily looking for you to solve the problem immediately; however, if you choose a solution, do be prepared to back it up in great detail.

Lean on your own expertise and knowledge to present what you know and walk interviewers through your thought process. Don’t be afraid to ask questions or explain areas of the solution that would require help from other people on the team. And don’t be intimidated to ask for help during the interview itself. “We structure our interviews to be collaborative so in the event that candidates are unsure, the interviewers can help coach them to a solution,” Butters says.


7 Questions You’re Likely to Get in Any Data Science Interview (and How to Answer Them)

Technical questions should be expected, but they’ll range broadly depending on what role you’re applying for and what your past experience has been. However, according to industry experts, recruiters, and hiring managers in the data science field, you will almost certainly be asked about the projects you’ve been working on, no matter what job you’re applying for.

Below, we’re providing some questions you’re likely to get in any data science interview along with some advice on what employers are looking for in your answers. (And remember that whatever job you’re interviewing for in any field, you should also be ready to answer these common interview questions.)


1. Can You Tell Me About a Recent Project You’ve Been Working on?

Hiring managers can infer a lot based on the way you talk about your projects, both past and present. Briana Brownell, Founder and CEO of Pure Strategy Inc., always asks about projects the candidate has worked on in order to assess the following qualities:

Communication skills: You should be able to explain in an accessible way what you did and why and outline some of the decisions you made as you worked on the project. What was the problem you were looking to solve? What were the solutions you considered? Make sure to touch on data sources you’re using, who your users are going to be, and how they’re going to use the information. Talk the interviewer through your thought process and do it in a way that shows you can translate technical processes into everyday language.

It might be helpful to talk a friend through your answer before the interview so you have practice communicating to someone who may not have the same technical background as you. A big part of your role as data scientist will be to bridge the gap between technical and non-technical staff, so the more comfortable you are when it comes to talking to “laypeople,” the better.

Teamwork: So much of the work you do in data science is collaborative, Brownell says, which makes it imperative that you can successfully work with others on a team. The hiring manager will be looking to see how you function as an individual within a cross-functional team, so make sure you weave information about who you were working with and how into your answer.

You may have worked in pair-programming situations that are highly collaborative or you could’ve been assigned your set of tasks and then had to come back together with the group. If the project you’re describing involved work with both technical and non-technical teammates, make sure to touch on that as well. (P.S. This may go without saying, but don’t throw your teammates under the bus in your answers!)

Enthusiasm: Don’t be scared to show off your excitement about your projects. For you, that could mean letting your sense of humor show as you’re talking about the project or maybe it’s using body language to emphasize the points you’re making or even explicitly mentioning why the work was so significant to you. When Brownell sees someone who is truly excited about a project they’ve completed, she considers it a huge plus.

A strong moral compass: Brownell is extremely impressed when she meets candidates who’ve considered the societal and/or ethical implications of their work. “I believe that this is one of the most important, growing areas in data science and will soon become a part of every data science process. At the senior level, data scientists should absolutely be thinking about the impacts of their work,” she says. If you can talk about how you considered the biases of the data set you used and the way certain decisions could have negatively impacted specific user groups, that’s going to be another plus.


2. Can You Break Down an Algorithm You Used on a Recent Project?

Maddie Shang, Machine Learning Researcher and Global Lead of AI and Data Science at Women Who Code, asks her candidates to select an algorithm they’ve used in the course of a recent project and then dives deeper with follow-up questions such as:

  • When should you use this algorithm?
  • When should you not use it?
  • Can you compare and contrast this algorithm with other similar ones? Why did you select this one instead of the others?
  • What are the underlying assumptions here? How did these assumptions respect or violate the data? How did you verify that?
  • What parameters and hyperparameters did you select and optimize? What did each hyperparameter do for the model? Did you have a separate parameter tuning data set (that was not included in the training and testing set)?
  • How does the algorithm scale with more data? With imbalanced data?
  • What is the run-time complexity and memory complexity of this algorithm?
  • Are you happy with the results? If I gave you more time, what might you have done to improve?

These questions will require you to reflect a lot on your data and project before the interview. So the first thing you need to do is think about which data and project you might want to talk about at this particular interview. If you choose an algorithm that’s related to the company’s work, that’s going to be really interesting to the hiring manager. If you have a good example that’s within your area of expertise, absolutely go for it!

Once you’ve chosen which project, data, and algorithms you’d like to talk about, it might be helpful to make a list or chart out all of the models that you tried and the associated analysis you did along the way. Shang points out that it’s best to start simple; if you can show that you began with very simple models, like GBDT models, and tried to overfit them on a small balanced sample, she would be very pleased with that answer. If, however, you went straight for a complex model, that would be a red flag to her that you may have a tendency to overcomplicate things.

When you’re talking about your results, the hiring manager will want to hear about the impact. If you’re talking about a personal project, still talk about the impact as if it was work done for a company or client. How did (or would) you communicate your findings to colleagues across the company or to a client? What did (or would) the deployment process look like? For example, it’s common to end up with a few models to choose from but to ultimately deploy the simpler one due to client preference; talking about this shows you understand the big picture.


3. What Tools Did You Use in a Recent Project and Why?

Nathan Oakley, Data Science Talent Acquisition Specialist at Syntelli Solutions, believes that companies always want to know you understand the pros and cons of various software, tools, and technologies and that you can make informed decisions about which ones are appropriate for any given situation. The alternative—that you’re simply using certain ones because that’s what you were told to do, and not for any other sound reason—would cast serious doubt on your abilities.

So use this question to show that you not only have a rationale for choosing specific technologies over others, but also have the communication skills to explain your decision. Think about all of the technologies you considered during your project. What were the reasons you decided to go with one over the other? Was there a trade-off you had to make but you felt was worth it? Walk the interviewer through your thought process, demonstrating how you weighed the benefits and drawbacks of various tools and which considerations won out and why.

Again, if you can tie any of the technologies you’ve used or considered to ones you know they use at the company you’re interviewing at, that’s great. Many times, you can find this information by searching online but if you’re unsure of what tech the company uses, that’s a great question to ask. Then, if you find out they use Azure, for example, and you know a bit about it, you can try to find ways to weave your experience with the tool and similarities between your past projects and ones you’d be working on in the new role into your answers whenever possible.


4. How Did You Help Solve the Problem Your Recent Project Presented?

You will probably get some questions around the solution you chose to your problem, predicts Oakley. These may be questions like:

  • What was the solution to the problem?
  • What model did you choose and why?
  • How did you arrive at this solution?
  • Was this solution given to you to execute?
  • Were you the one who identified and/or designed the solution?
  • What is the impact the solution will have on the company?

To start composing your answer, ask yourself: How did I add to the success of this project? Provide specific examples that show you have an understanding of what your purpose is within the bigger picture of the team you were on—showing that you can be creative about problem solving and collaborate effectively.

Maybe you had an idea that ended up directing the team in a completely different direction, or perhaps you were able to offer a unique perspective when it came to a specific methodology because you had experience using it previously and could warn teammates of some of the unanticipated drawbacks. Drill down to identify your contribution to the solution—and your explanation of how the solution affects your client or company.


5. Tell Me About a Challenge You Faced During the Course of the Project.

Any employer will want to know that you can successfully handle change and challenges. Oakley says that questions asked in this realm can include:

  • What challenges came up during this project?
  • Were your challenges primarily related to data, modeling, internal communication, organizational roadblocks, or something else?
  • How did you work to overcome these challenges?

When you’re talking about the obstacles you encountered, what the interviewer wants to know more than anything is how you reacted and what you did to fix or address the situation.

For example, maybe your tendency is to jump up and take a leadership role when it comes to solutioning (which would be something critical to show if you were interviewing for a lead role). Or perhaps you’re more junior, but you still found a way to make your voice heard or you played a crucial role figuring out how to practically implement a solution the team came up with together. Or perhaps the challenge was more about a miscommunication you had with another team about the goals or timeline of a project, and you took the lead as a liaison to arrange a much-needed conversation among the various stakeholders.

No matter which challenge you choose to discuss, the best way to frame your answer to this kind of behavioral interview question is to use the STAR method:

  • Situation: Give a brief explanation of what the circumstances were.
  • Task: Explain what your role in the matter was.
  • Action: Talk through the actions you took in order to address the issue.
  • Result: Don’t forget to tell the interviewer what the results of your actions were—in terms of straightforward project output as well as learnings and insights gleaned for you and other stakeholders.


6. Looking Back, If You Could Do One Thing Differently or Improve Just One Thing on the Project, What Would It Be?

Dosad refers to this as the “pull-apart” question. Hiring managers want you to talk them through a previous project, but through the lens of your current knowledge. “They’ll want to see how you would have changed the work that you did previously and what you would do now under different circumstances,” he explains.

The overarching point of this question is to assess your desire to continue improving. “At the heart of it all, data scientists should be inquisitive and constantly be striving for new discoveries and new areas of efficiency and accuracy,” says Dosad. “Candidates who are unwilling or unable to critically assess the work they have done in the past will typically not make it past this interview question.”

If you need a place to start when it comes to answering this question, think about the challenges you’ve already outlined for the question above. Could any of those have been prevented by doing something differently before you got to that point in the process? Knowing everything you know now, what would you change?

Be honest about what you’ve learned and how you’ve grown as a data scientist, as a teammate, and/or as a leader. But do it tactfully; it’s never a good look to badmouth a former company or any of your colleagues during an interview. In other words, your answer should be about how you’ve changed and see things differently, not about how you wish other people or things hadn’t messed up.


7. Let’s Refer Back to XYZ Technical Problem You Solved. If You Had an Infinite Budget, What Would Be the First Thing You’d Do to Fix That Same Problem?

While there are “many schools of thought on how to answer this, the safest answer typically is around gaining more access to data,” Dosad says. “A lot of data scientists will typically think about building neural networks or getting state-of-the-art GPUs [graphics processing unit], but in this instance, the more data you have access to, the more accurate you can train a model to be. It may not be the most interesting answer, but it’s typically the one that will yield the best results.”

Interviewers want to see how you approach problem solving. If you jump to the most common answer—buying more equipment or going after the largest and most complex solution out there—that shows a limited understanding of how problems in data science work. In this field, the biggest component is data and accessing it. You should answer accordingly by talking about exactly which other data you could have accessed and why, explaining how specifically that additional data would have led to better results in the end.