The world’s leading publication for data science, AI, and ML professionals.

The Non-Company Specific Approach to Data Science Interview Preparation

Understanding what you need to know to land a job as a data scientist

DATA SCIENCE INTERVIEW

Photo by Van Tay Media on Unsplash
Photo by Van Tay Media on Unsplash

If you have ever interviewed for a Data Science position, you know that it takes a lot more than passing a single interview. There are phone screenings and multiple rounds of on-site interviews to get through. Whether you already have interviews or are just working to prepare, it’s overwhelming just thinking about everything you need to know.

And what exactly do you need to know? In this post, we are going to take some of the scariness and uncertainty out of interview preparation. We will look at the different types of interviews you will encounter when pursuing a position in analytics driven and algorithm driven data scientist roles and the non-company specific approach for preparing efficiently and effectively, which is what I recommend. Make sure to read to the end so that you don’t miss out on why this method works!

Let’s get started!

Table of Contents

The Types of Interviews for Data Scientists

When interviewing for a data scientist role at a tech company, you need to be prepared to face multiple types of interviews, so the first critical part of preparation is knowing what interviews you should expect and what the interviewer expects from you in each of those interviews.

Before we dive into the different types, let’s first consider the process a bit more. Once you make it to the interview stage for a data scientist position, what can you expect? The interview rounds will typically consist of 1 to 2 rounds of technical phone screening and 4 to 6 rounds of onsite interviews. Some companies also have online assessments and take-home assignments in the interview process.

Photo by Glenn Carstens-Peters on Unsplash
Photo by Glenn Carstens-Peters on Unsplash

This is why you need to be familiar with all six types of interviews I’m about to outline. Although, you likely won’t encounter all of them, if you want to make it through all the interviews and land the job, you are going to need to pass several interviews and that takes knowing what to expect.

So, as we go through the types of interviews, we will look at how each of these interviews is designed and how they are used to evaluate candidates.

The Product Case Interview

The Product Case Interview, which is also sometimes called the metric interview or the business case interview, evaluates your ability to measure, and your knowledge on product and a/b testing, which are core skills of data scientists, especially analytics-driven data scientists. So the product case interview is designed to test your product knowledge and critical thinking skills.

In the product case interview, you are given a business scenario or problem and then asked to explain your approach to solve the problem and make suggestions. The questions can vary quite a bit, but they might include things like:

  • Figuring out why a metric has declined
  • Designing experiments
  • Making decisions about whether to launch a product.

Some example interview questions are:

  • Can you provide some metrics to measure user engagement?
  • What are the pros and cons of using them?
  • What are some things to consider when running an experiment?

To do well in this interview, you will need to demonstrate an ability to design metrics, diagnose metric shifts, and have ideas about how to improve a product and solve business problems. To learn more about this type of interview and how to prepare, I recommend checking out this whole playlist on cracking this kind of interview, which includes not only tips but also sample answers to real interview questions.

SQL Interview

The SQL Interview is critical as SQL is used very often in the daily job of a data scientist. Whether you work for a small or a big company, you will use this tool consistently to do data analysis, diagnose issues, and gather insights from data. Therefore, you must demonstrate proficiency in this interview.

The SQL interview will test your familiarity with SQL language, syntax, and function, but that isn’t all. It will also test your ability to logically understand the problem and create efficient queries in a short amount of time. The questions cover things like:

  • Filtering data
  • Joining different tables
  • Computing complex business metrics on users activities or system logs

I’m not going to give sample questions for this type of interview because you can find many sample questions on Leetcode and Hackerrank.

Probability and Statistics Interview

The Probability and Statistics Interview is as technical as it gets. The questions will test your knowledge of applied statistics and probability. Applied statistics is particularly important as a data scientist. It helps you with leading A/B testing, doing data analysis, and making data-driven decisions. Thus, the probability and statistics interview is where you must demonstrate that you have the technical skills necessary to perform the job.

The question in this type of interview can be conceptual such as:

  • What is p-value?
  • How do you calculate confidence intervals?
  • Can you explain the central limit theorem?
  • Can you list the assumptions of linear regression?

They can also involve calculations, such as:

  • What is the probability of obtaining 2 consecutive heads from 5 coin tosses?
  • What’s the probability of winning a game given some specific conditions?

If you want to get a better sense of statistics interview questions, feel free to check out another playlist on my YouTube channel where I go over some commonly asked questions and answers.

The Machine Learning Interview

The Machine Learning Interview is closely related to the responsibilities in an algorithm driven position. You must demonstrate your knowledge of the basics of machine learning. The questions will look at things like:

  • How machine learning models work
  • The pros and cons of different models
  • Techniques that are applied when dealing with different datasets
  • Tuning hyperparameters

Example questions for this type of interview include:

  • What is the overfitting problem?
  • How do you deal with it?
  • How to deal with an imbalanced dataset?
  • What does Random in Random Forest Model mean?
  • What are some metrics to evaluate a classification model?

For more information on machine learning interviews, I recommend this blog by me and my friend Zi, who is a machine learning engineer at Google. My channel also has a playlist about machine learning interviews which covers various topics from an overview of the machine learning interview to the implementation of commonly asked algorithms.

The Coding Interview

The Coding Interview tests your coding ability, which includes proficiency in programming or scripting languages such as Python. Additionally, you will need to show that you possess computer science fundamentals, including an understanding of algorithms and data structures.

Some people consider coding to be the same as SQL, but it’s actually not. The coding interview refers to coding up algorithms, like binary search, quick select and data structures, like list, stack, queue, tree and graph, etc. The questions vary from implementing a simple algorithm, such as a quick select, to solving a more complicated problem involving using a search algorithm, such as BFS or DFS, on a tree or a graph data structure.

In this interview, you are expected to have a logical understanding of problems and be able to come up with efficient solutions within a limited amount of time. I will not give sample questions for this type of interview as most of you are familiar with LeetCode, where you can find many examples.

The Behavioral and Experience Interview

The last type of interview for data scientists is the Behavioral & Experience Interview. In this interview, **** you will be asked about what you would do in hypothetical situations and also about how you have worked with others in the past. You will also be asked about past teams and projects. The goal is to make sure that you fit with the company culture.

As a candidate, it is crucial that you don’t overlook this interview, thinking that it is not as important. They are most certainly important! Big companies such as Google and Amazon are putting more emphasis on these interviews. Some example questions include:

  • Can you talk about a previous data science project you have done?
  • What problems have you solved at your current or previous job, and how did it contribute to the overall success of the company?
  • Can you tell me a time you went above and beyond your responsibilities?
  • What are some of the difficulties you have faced in the past with your work? How did you resolve them?

If you want to learn more about behavioral interviews, don’t worry, I have you covered. I also have a playlist on behavioral interviews including how to present past projects and the dos and don’ts in a behavioral interview.

What Interviews Should I Focus On?

We’ve covered six types of interviews, which is quite a lot to keep track of. At this point, it’s helpful to remember that the exact type of position you are targeting will make a difference in which interviews you should focus on. If you are unsure what position you should be targeting, then I highly recommend reading this blog post on Targeting the Right Data Science Position for You.

Although they are both data scientists, analytics driven and algorithm driven roles do differ and require different skills, and the interview process will reflect this. For the analytics driven roles, you should prioritize the product case and SQL interviews. For the algorithm driven roles, you should focus on coding and machine learning. The probability and statistics interview and the behavior and experience interview will be important for both tracks.

The Benefits of a Non-Company Specific Approach

So those are the types of interviews, and at this point, you may be starting to feel very overwhelmed. Even when you consider what position you are targeting, there is still a lot of interview preparation to do. How should you approach such a large task?

The interview preparation method I recommend is a non-company specific approach. A non-company specific approach means focusing on building a strong foundation and working on the fundamentals like statistics, product sense, and a/b testing first before approaching interview questions.

To understand why I recommend the non-company specific approach, it helps to look more closely at the alternative, which would be a company-specific approach. Many people prepare for interviews this way. When they get an interview for a specific company, they look up sample interview questions online and prepare for the interview that way.

Why do I not recommend a company-specific approach? The first reason is that you don’t have time to get fully ready. If you wait until you actually have interviews to start preparing, you are going to have a lot to cover in a very limited time, especially if you have interviews with more than one company. There are a lot of interview types, so if you start only once you have an interview, you are going to be doing a lot of cramming. You likely won’t be able to study everything, and you won’t have time to study anything really well.

Sometimes, I have students sharing this kind of experience with me – "I cannot believe I failed the interview because of a simple SQL question or because of a simple stats question."

When we look into why that happened, it typically turns out that they spent all the time focusing on collecting company-specific questions online, hoping to get the exact same questions during the real interview. What ended up happening was that they did not have enough time to brush up on the basics, nor did they get the interview questions they prepared.

Another reason I don’t recommend a company specific approach is that it forces you to start from scratch with every interview, and it leaves you with a rigid rather than a flexible working knowledge.

If your preparation is focused on learning the answers to questions you find online, then you may not be unable to adapt to questions that differ from what you studied. This makes it especially difficult to answer follow-up questions. You will also be out of luck if you cannot find questions online or if the interviewer just happens to ask different questions than what you found, which is a very real possibility.

In contrast, a non-company specific approach gives you a deeper understanding and is actually more efficient. You can and should start BEFORE you land an interview which means more overall time to prepare, and if you have multiple interviews approaching, you can prepare for them all simultaneously. Considering how many types of interviews there are, being able to prepare efficiently is hugely important.

Photo by Brooke Cagle on Unsplash
Photo by Brooke Cagle on Unsplash

A non-company specific approach is even more valuable though for the type of knowledge it gives you. When you focus on the fundamentals rather than specific questions, you are truly learning rather than memorizing.

This type of approach forces you to become far more comfortable and confident with the information. When you learn things this way, you won’t have to relearn them again for every interview, again making this method more efficient.

So, as you start your interview preparation, whether you have landed your first interview or not, I recommend a more general approach. It can feel like there is far too much information to study this way, but in the long run, it will actually save you time and help you be better prepared.

Conclusion

That’s everything for this post! Remember that preparing for interviews involves knowing what to prepare for and being proactive with your studying. Don’t wait around till you land an interview and then try to cram! Get started studying now!

Speaking of studying, with six types of interviews to prepare for, you might be wondering how you can make the most of your study time. If so, be sure to check out this blog which will give you study tips to improve your interview preparation.


Thanks for Reading!

If you like this post and want to support me…

Why Aren’t You Getting Data Science Interviews?

Cracking Statistics Interviews for Data Scientists


Related Articles