10 Data Science Questions You Should Ask Your Candidate in an Interview (And How to Answer Them)

Data science is one of the fastest growing fields in the world, with almost every company expanding their data analytics team.There’s so many opportunities for you out there.

That being said, the interview processes are only getting harder.To help you out, I’ve compiled a list of 10 questions that are likely to come up in interviews, and how to approach them.

For me as a data science hiring manager it’s important to be able to filter out unsuitable candidates quickly. A good interview won’t be full of technical question but will test the core competencies ideal in a data scientist.

So, what are the most important data science interview questions?

These questions are designed to test a candidate’s general aptitude for data science as well as their understanding of business needs. A good data scientist at every level needs that balance of technical and business understanding, coupled with great soft skills.

With all that in mind, let’s jump straight into the top ten questions that you’ll want to prepare for. We’ve split each question into what it means for the interviewer, and what you should think about as an interviewee.

Tell me about a piece of data science research that you have read recently?

Interviewer:

This question can tell you a lot about how much a candidate is interested in the field in general. It also gives you a better understanding of the type of work that interests them. Follow up questions should include how they can apply the knowledge to your business.

Candidate:

Have a go to piece of research that you are able to talk about for roughly five minutes and explain in some detail what the conclusions mean for the community and the paper. You should also be able to explain how at least some of the technical aspects work. If you’ve tried to replicate the results yourself and can talk about your experience, that’s even better

What do you think are the key skills for a data scientist? 

Interviewer:

This is a great way to get the candidate’s thoughts and views on what it means to be a good data scientist. It will also give you an idea of the type of data scientist they are likely to be. 

For example, if they answer that it is important to keep up to date on the latest research then you know that they will be a very technical data scientist. 

Candidate:

First and foremost, be honest. You want the hiring manager to align your skills with the skills already in the team. So if you aren’t interested in the stakeholder management parts of the role then dont talk much about them. It’s a two-way interview remember, you want the job that suits you just as much as they want an employee to suit their role.

What are the key stages in any modelling project?

Interviewer:

I like asking this one. You can always tell who has had more involvement with the entire project with this simple question.

If a candidate is quite junior they are likely to only talk about the data phases: sourcing, cleaning, modelling. However, a more senior data scientist will talk about the business hypothesis to be solved, how the model or analysis will be implemented and so on. 

Candidate:

As a candidate, you will want to think more broadly about the project. Why is it being undertaken, what is the stakeholder expectation, how much effort will it be to put into production?

How much time is spent on each stage of a project?

Interviewer:

This could be a follow up to the previous question. You want to get an understanding of how they think about the whole process, not just their part.

Candidate

As a candidate, you need to show that you realise how your work fits into the rest of the team. Talk about how your work sits alongside others, and the needs of working with stakeholders, other technical teams.

What are common issues with data sets that the business generates?

Interviewer:

Here, you want to make sure that the candidate understands some of the issues they may face and how they can clean the data for use. Common issues include: poor data coverage, inconsistent typing, inconsistent collection or data entry, data inconsistency due to legacy systems.

Candidate:

You need to be able to identify 3 or 4 common issues and propose how to deal with those issues. For example, can you discuss how to deal with null values in numerical data? It is also important to understand how the data changes over time and if that affects any static models or metadata.

What is your ideal tech stack for data science?

Interviewer:

This is very open ended. There is no 100% correct answer here but you can learn what a candidate likes to work in. It’s great to just have a conversation with your prospective employee to get an understanding for how they like to work.

Candidate:

This is your chance to show what you’re comfortable using and see if your interests and expertise line up with the company’s data science workstack. Be honest, but also bear in mind that the type of role will dictate to some degree what technology you use. Ask questions! It’s great to show curiosity. If they use something you don’t would you be keen to learn?

What would the distribution of chicken sizes in the USA look like?

Interviewer:

This question should test the candidate’s critical thinking. The actual answer isn’t important here, what is important is that a candidate can reason through the question and provide and challenge assumptions. Pay close attention to how they think about solving it and encourage them to speak out loud during their thought process.

Candidate:

Think out loud; Show your working to the interviewer and talk through your thoughts. Ensuring that the interviewer knows what you are thinking every step of the way is the most important thing. If you are making an assumption then state that assumption.

What are the biggest blockers for good data science work in a business setting?

Interviewer:

This is an opportunity to get the candidate thinking about how they can fix the blockers as well as identify them. Common answers are: poor data governance, poor technology, unrealistic expectations. Be positive when you’re asking these questions, try and prompt some follow up thoughts.

Candidate:

Look back on your previous roles and use your experience! What were the most annoying parts of producing good data science work, what could’ve been improved? Why wasn’t it improved?

Again, think aloud, allow the interviewer an opportunity to see how you think.

What is the most interesting project you have worked on in the past year?

Interviewer:

This is a pretty standard question but it is still incredibly insightful. It’s not just about the answer, it’s just as much about how they reply. You want to hear the passion in their voice when they talk about their work, whether it was simple analysis or constructing a whole new process for data cleansing. 

Candidate:

If you can’t list a top five of your favourite projects on command, then practice this question.You may not have had great experiences, but you need to find inspiration, and really be able to convey that to your interviewer.

Tip: list all the projects you have worked on in your career from most interesting to least. Discard all but the top five. Rank these in terms of how much you were involved in the project. The project at the top should be the one you talk about. If it’s a mix of positives, negatives that were learning points, outcomes, then you have a perfect discussion point!

What makes a good data science manager/ team lead?

Interviewer:

This might be the most important question as a hiring manager. Their answer should show you whether or not you would be a good fit for them, and whether or not there could be clashes. You want to be honest with yourself as well. Use it as a learning point whilst also trying to understand how they like to be managed.

Candidate:

Think about what you like in a manager. Do you only want them there to put out fires or do you like having someone hands on? Make sure you research the company’s general approach to leadership. Align your answer with their culture, but also remember that it’s in your interests to be honest. You want a manager that can work with you too!

Wrapping up

These questions are designed to get a candidate thinking about more than just statistics, coding and machine learning. They tell you a lot more about how the candidate approaches problems, what their expectations of a data science role are and their strengths and weaknesses.

Bear in mind there are no right or wrong answers for a lot of these questions. They are made to test critical thinking, business fit and passion for data science.

At its core, data science is a technical field. Pinata prides itself on our technical articles coupled with theory to get the most out of our learning. Start here.

If you want more career advice then consider subscribing for our email blast where we share tips and tricks to maximise your chance of getting a new job!