On Interview Query, you’ll find plenty of resources on WHAT to study for data science interviews. But we haven’t written a lot about HOW to prepare for the data science science interview.
After talking with hundreds of data scientists who have landed jobs at FAANG companies and other startups, we’ve landed on a simple study plan for data science interviews. These four steps will help you prepare for the interview and manage your study time:
If you have a data science interview coming up, follow these steps on how to prepare for the data science interview. This process will help you study more effectively, build your confidence in interviews, and help you simulate the interview process.
Data science is a HUGE domain, and because of that, studying the right data science interview questions is half the battle. Somehow, I still see candidates mess this up every time.
They’ll religiously study machine learning concepts, only to get slapped in the face with a business case question during their technical screen.
Or, they’ll review a ton of case studies, only to be asked about matrix multiplication in numpy.
Preventing yourself from making this mistake is usually easily avoidable with some due diligence. To review, data science interviews consist of around 10+ different interview topics. Ordered alphabetically, they are:
So, given this wide breadth, how do you study everything?
Simply put, you don’t. Here’s what you do instead:
Instead, if you actually have the interview scheduled, reach out to the recruiter and ask them what kind of questions will be on the interview. PSA: They want to help you and see you succeed! I cannot stress how easy and beneficial this task is.
In the worst case, they say no. Best case, they tell you exactly what’s going to happen in the interview because, newsflash, they get paid when you get the job. Remember, the goal is to narrow the breadth of topics that you have to study, allowing you to go deeper into the subjects that matter and be more well prepared.
If you’re not at the interview stage yet, you can start preparing by doing research, reading job descriptions, and finding interview experiences from other candidates on Blind or Interview Query. Look out for common subjects, like if they talk about A/B testing or Python experience in their roles and responsibilities.
At Interview Query, we provide an exhaustive list of company guides where we’ve already analyzed which topics are asked for different roles. You can browse our interview guides here. We also have in-depth guides for companies like Meta and Google.
Once you have a topic list, you should focus on understanding your familiarity with those question topics. Do this by just creating a list of question topics that’ll be asked in each interview you’re preparing for, and order it by the most important across all interviews.
For example, if I’m studying for the Google data scientist interview, I would first prioritize statistics and A/B testing and algorithms, and then work on machine learning and SQL questions next, with some probability questions to round it all out.
Practice a few different levels of statistics interview questions to understand your knowledge. If you find yourself easily solving the easy and medium questions, then you should be confident to move onto another subject where the questions are a little harder to grasp.
The goal is to have a list of question topics with your confidence level next to each one. This way, you’ll have an idea of which subject you need to spend the most time learning, practicing, and reviewing.
James Clear once said: “In the gym, if you experience no stimulus, your muscles won’t grow. If you step under 10,000 pounds, your body will break.”
Studying for data science interviews is very similar.
We are in the business of building brain muscles to solve very specific problems. Practicing bite-sized interview questions on a continual basis allows you to level up slowly and build on top of your existing knowledge. You can only solve medium level questions once you can ace easy questions, and so on for hard questions.
Additionally, one thing we’ve learned at Interview Query is that the more problems you study, the more likely you are to pass an interview. This intuitively makes sense, but it’s also backed up by data.
We looked at exit survey data from members on Interview Query and saw a correlation in goal achievement and the number of questions studied:
I will add a slight caveat that, as a data scientist, this is slightly biased. You would expect that candidates who study a lot of questions are probably more diligent in other areas of their interview preparation (or life) as well.
But this is still a good representation of how important maintaining effort is towards achieving success. And I’m not going to lie, if anyone studied over 100 interview questions, you would think they’re probably going to do well on their interview.
So, building an interview study habit is easier than it sounds. My advice is to try two specific things:
Most of the time, it takes putting in small steps. That’s why it’s important to start out by solving some easy interview questions first or just trying a few multiple choice questions. Maybe start doing 5 minutes of just thinking about a problem.
The goal is to build a habit and the best way to do so is to make the first steps fairly easy. Then, as you get used to the routine, you can steadily work up the difficulty.
Studies show that nudges towards your goals are great resources towards building positive habits. Just like how companies like Facebook send you notifications to get you addicted to Instagram, you can also do that with studying to help you achieve your goals!
If you sign up for Interview Query, we’ll send you one question per week in your email for you to practice.
It’s hard to grind and study for your data science interviews without getting feedback. Continual feedback is what makes doing tasks much easier because we’re getting recognition or insights into the work you’re doing instead of it feeling like a slog.
Mock interviews and coaching are great simulations for actually testing your live performance against another person. If that person is specifically a data science coach, they can dive into your process and explain exactly how you can improve in certain areas and further develop the skills to effectively communicate your ideas and solutions
Additionally, many interviews aren’t like practice problems because they include problems that slowly build on top of your existing answers. If you answer a case question on your own, it’s a lot harder to dive in-depth.
So, by being forced to clarify questions and walk interviewers through different solutions, you’ll build up the mental practice to solve problems in real time.
If you’d like more tips, please subscribe to my channel, like this video, and I hope that you can stick to these methods.
It’s no secret that, in any interview, companies are always interviewing multiple candidates for the role. Sometimes you can get the job by just passing their expectations if they’re more desperate for hires, but most often, you have to be the best interviewing candidate out of a pool of candidates.
This happens when you have experienced interviewers, who in their head have a distribution of skill level from asking the same question multiple times and understanding what a good answer sounds like versus what a bad answer sounds like.