The Data Science Interview: What to Expect

The data science interview is feared because of its uncertainty. Here, I break down the process and include tips to help you succeed

Across many industries, the interview process is something that people fear. In consulting and banking, prospects study for countless hours to master case interviews and technical questions. For software engineers, an entire new business model has developed purely based on helping prospects answer coding questions. Many of the fears around interviewing come from uncertainty about the process. In data science, which is a relatively new field, this is shrouded in even more mystery.

All companies interview for data science differently; however, there are a few similarities that you should expect to see in most of the jobs that you apply for. In this article, I do my best to break down the general interview phases and give some tips on how you can excel during each of them. Hopefully this will relieve some of the stress associated with the uncertainty in data science interviews.

For the purpose of this article, we will skip the nuances of the application process. In the future, I plan to write an article about how to optimize your resume and your communication channels with companies, so stay tuned!

Phase 1: The Phone Interview (15–30 min)

For most companies, the first step is to talk to you on the phone. You usually talk with either a technical recruiter (large companies) or a data science manager (smaller companies).

These conversations are short and are primarily focused on understanding your fit or your past project experience. Occasionally, they will ask for a broad assessment of your relevant skills.

Tips to Succeed:

• Think carefully about why you want to work for the company. Recruiters get excited about passion for the company’s mission or excitement about the specific role you are applying for
• Review the job posting and make sure that you can speak to your experience with any of the relevant tools
• Review your resume and make sure that you can speak clearly about any of the past projects you have worked on. For bonus points, tie these projects back to the work that you would potentially be doing on the job
• This is a good opportunity to ask questions of your own. Asking thoughtful questions that show you researched the company thoroughly can improve your odds of getting to the next round

Phase 2: The Take Home Test

When you get past the phone interview, companies often send you a take home assessment. This can be a dataset for you to analyze, a coding assessment, or a project that they would like you to present on. This will vary greatly by the company and the role.

Data Analysis: I personally believe this option is most representative of the work that you will be doing on a daily basis. For this type of question, they will usually ask you to complete the analysis in a relatively short time frame. They will say something like: “This shouldn’t take you no more than 3 hours”. From my experience, they almost always take significantly longer than that. I would try to set it up so that you can perform this over the weekend if possible.

Tips to Succeed:

• If the don’t explicitly say you can’t bring in data from outside sources, consider appending other useful data.
• Using feature engineering or feature reduction (and being able to explain why) can improve the quality of your analysis greatly
• Putting some time to making your work interpretable to business stakeholders (detailed visuals) can show that you understand how to drive business value
• Creating an API endpoint or web app illustrates a bias towards actionability in your work (this may be overkill)

Coding Assessment: Unless you are applying for a machine learning engineer role, coding assessments are generally on the easier end for data science positions. You should expect that they will let you choose your preferred programming language to answer a few questions.

For these, there is generally a strict time limit. They either have you log on with an interviewer who will watch you perform the assessment, or they will have you take the test through an online platform unmonitored.

Honestly, these even make me nervous even if they are relatively simple. There is something about time pressure when coding or problem solving that is inherently uncomfortable.

Tips to Succeed:

• Practice, practice, practice. There are plenty of places online to practice coding questions. The best known one is probably leetcode
• Go on glassdoor.com to see if you can get a feeling for what some of the questions may be like
• If you are doing a assessment where the interviewer is there, practice talking your way through your code. It helps to actually have someone there watching you, even if they don’t know anything about coding or data science

Presentation: These are quite rare. Some companies may ask you to present on a past project or to run a simple analysis and present your findings via powerpoint. If they are making you present, they are likely trying to test your ability to speak to business stakeholders.

Tips to Succeed:

• Try to focus on telling a good story about your project, why you used the methods that you did, and what the outcomes were
• Take time to make your slides visually appealing
• Back up every statement you make with data
• Prepare for potential questions beforehand

If you make it past the first two phases, they will bring you to the office for an in-person interview. There are exceptions to this if you are living across the country or are applying for a remote job (video interviews).

When in the office, you will usually do 2–5 interviews with people on the team and across the company. The following phases can come in any order. Your interview schedule is largely dependent on the schedules of the people interviewing you.

Phase 3: Interview with a Data Scientist (30 min — 1 hr)

After you come into the office, companies will have you speak with one of the data scientists on the team. This interview can get fairly technical. Expect them to ask about the projects on your resume and the logic behind your methodology. They may also ask you to explain some of the math behind the algorithms that you used.

If they have any questions about the take home test, they will generally ask them here as well. Since this is a project that they are familiar with (they may have taken the assessment themselves), be prepared for questions about the data cleaning, model choice, etc.

Finally, if there is time, they may ask a few behavioral interview questions. These are important because if you get the job, you will likely be working with this person. If you can create a connection here, this can go a long way.

Tips for Success:

• Take time to review the projects on your resume thoroughly. Be able to explain any of the algorithms that you used in depth
• Review general data science interview questions, these are generally statistics based. You can find examples of these here
• Review your assessment and think about ways that you could have improved the analysis
• Ask questions about his/her experience with the company

Phase 4: On site coding exam (30 min — 1 hr)

Often, the same data scientist in the previous section will give you another quick technical assessment. This one primarily focuses on SQL or some light python data manipulation.

For these questions they will either give you a computer, or ask you to solve the problem on a white board. They will expect you to talk your way through your code.

Tips for Success:

• Practice SQL questions online. Make sure you understand query basics (SELECT, FROM, WHERE, GROUPBY, HAVING, ORDERBY), types of joins, self joins, subqueries, etc.
• Get comfortable with pandas and numpy for data manipulation (if Python can be used)
• Practice solving questions on a piece of paper or on a whiteboard with a partner present

Phase 5: Interview with Data science manager or lead (30 min — 1 hr)

At some point during the day, you will speak with one of the people managers. This person can be a Director of Data Science, VP of Analytics, Data Science Manager, etc.. They will be focusing again on your fit for the team.

Expect them to ask about the past projects you have worked on. Their focus will generally be more on the process you used rather than on the technical algorithms. They may also ask about your experience with project managers and different PM philosophies. Technical questions are generally fairly rare in these circumstances unless this manager has a heavily quantitative background.

In the behavioral portion of this interview, they would like to better understand your goals. They may ask about how you would like to grow in the company, or what you would like to learn on the job.

Tips for Success:

• Ask in one of the earlier stages about the PM philosophy that the team uses. Do some homework so that you have at least a basic understanding of it
• Think about what you would like to learn on the job and what type of trajectory you would like to have in the company
• Ask questions about the management philosophy, work autonomy, and the career development that the company offers

Phase 6: Debrief (30 min — 1 hr)

This interview is usually saved for the end. In larger companies this will be with a HR representative, but in some smaller companies this may be an opportunity to meet with the CTO or CEO. This conversation will also have a significant fit component. It is your opportunity to reiterate your excitement about the company and its mission, why you would be a good fit, and how you could create unique value to the organization.

From my experience, your interviewer will ask a few questions then let you take control.

Tips for Success:

• Don’t be afraid to talk about yourself as long as you relate it back to why you love the company or how you could create value for it
• Be complimentary, if you are talking to the CEO or CTO, it doesn’t hurt to say how much you enjoyed the other interviews
• Ask questions about the future of the company and next steps

Others You May Interview With

The main suspects will be members of the data science team or technical recruiters, but there are a few other people that could be a part of the interview process. There is also a chance that multiple people will interview you at the same time.

These are a few other types of interviewers:

• Project Managers — If you get the job, you will work with these people daily. They will usually ask you about your working style and your experience with different types of project management frameworks.
• Business Stakeholders — These will be the people that benefit from your work. They may ask about how you convey information and findings.
• Hiring Managers — This group is focused on your fit in the company. They want to see if you are passionate about the work and if you will buy in to the company culture.
• Data Engineers or Software Engineers — Sometimes they will have these people interview you if you will be working with them regularly. They may also administer the SQL assessment.

Final Thoughts

Hopefully this guide gives you a better understanding about what to expect in interviews. When there is less uncertainty about this process, I believe that it can actually be fun. If you want to succeed, it is important to excel in the behavioral portions as well as the technical ones. Many times I have seen candidates get jobs because interviewers believed in their potential and passion in spite of marginal technical ability. Before the interview, you should try to do as much research as possible about the people who will be interviewing you. Finding commonalities and trends in people is just as important as it is with data.