Cracking Data Science Interviews -Sharing My learnings from 25+ Interviews

A Framework to crack any Data Science Interview based on my learnings

Greetings and thank you for taking the time to read this article. Mu Sigma Inc. has been my employer since I began my data science journey as an intern more than 3.5 years ago. I intended to change jobs in order to gain more experience in this field. Then, despite having 10+ offers from top analytics / AI teams, I recently joined a startup as a Data Scientist II in the domain of Remote Sensing + Geospatial Artificial Intelligence.

Despite the abundance of articles on how to prepare for a data science interview, I want to share my interview experience, lessons I learned, and a few pieces of advice that I found useful. By creating a framework (which includes clear phases and milestones, checklists, and recommendations) to make your preparation and job search easier and more effective, I’m attempting to make this more structured because most of the articles are just a list of hazy themes or notions. Any IT or data professionals looking to change careers would benefit from this.

Welcome to my Pragmatic Data Science — 2023 Series

The framework consists of 4 major blocks —

1. Assess Yourself
2. Grab the Attention
3. Learn — ReLearn — UnLearn
4. Present Yourself

Let’s get started with a Detailed view of each block

Assess Yourself

It’s crucial to evaluate your skill set and fit with the market expectations before beginning your search for a job in data science (or any other field). I want to underline a few essential requirements that must be met because I notice many people wanting to join DS jobs. I am breaking down the self-assessment into three steps, like1. Observe your experience, 2. Assess yourself and your experience, 3. Define objectives

How to observe our experience —You’ve finally made up your mind to change, so let’s look at what you’ve done so far and see if it makes sense in terms of the market. Use this sample as a guide for this —access the Link here

With the help of the template, you may keep track of every significant experience milestone by answering the columns’ thorough questions. This will also be useful for writing your résumé and providing answers during an interview.

You should place a strong emphasis on the results of your projects, their scale, and the methods you used to complete them. Given that the answers to these questions are inextricably linked to more logical explanations, they will set you apart from profiles that are only loaded with empty phrases, phoney initiatives, and false claims.

Once you have the experience tracker, evaluate the skill sets listed there and determine whether they are still applicable today. You may possibly test yourself online and try to design an ML pipeline for benchmark problem statements.

Few Sources: Hackerrank and HackerEarth would help you to practice, assess, and showcase. ( These two would be sufficient for DA, Programming, and Basic ML, along with that You can evaluate yourself on a few Kaggle problems)

For example, If I had mentioned SQL as a skill that I used for a project, I need to evaluate my current rating for that. To be honest, in almost all the interviews I faced difficulties in solving SQL questions, even failing to clear a few because of my low SQL ratings.

Likewise , During my internship i have worked on many projects based on R , Rshiny , dplyr stack — but these days the relavancy of these skillsets are very less to the DS job descriptions. That’s why I keep insisting on having strong thought process to approach a problem rather just a memorized few lines of code.

Here I would like to reiterate “ when you prep for DS interviews — especially when you target DS2 / Data Science consultant / SDE2 equivalent roles, you should double check that you have enough Hands-on delivery experience in Advanced Data Analysis, Machine Learning, Deep Learning — Maybe you can clear interviews with many plan Bs or shortcuts, but when you start working on that role, It will really hit you from the beginning”.

As a result, having a precise self-assessment will assist you in determining what to learn and where to improve. Okay!… Now that you have a complete understanding of your abilities, you can decide what is feasible for you based on the percentage of matching. The map below can help you decide how to proceed with the next steps.

If you have negligible (I would say 50% is negligible) relevance to current DS roles and expected skill sets, you are not a Job Ready Profile, which means you must invest time in developing strong ML / DS foundations. The reason I say this is because “having a Data Scientist / Decision Scientist in your current job role name does not always mean that you can expect the same role in your switch.” It is always a good idea to compare ourselves to what the industry expects from a resource.

Most DS interviews I’ve had have had 4+ rounds, with a minimum of 2–3 rounds evaluating our technical skills, particularly on fundamentals. To pass those technical assignments, MCQs, or even one-on-one interviews (In a few interviews, my first round was with the Head of DS / Principal DS to assess my breadth of knowledge and ensure I met the basic expectations.)

Grab the Attention

You can confidently feel that you are “Hirable” at the end of the preceding Stage. The following three additional steps are critical to making your interview preparation results tangible. You must now make yourself visible to Recruiters

Resume

1. You will need a detailed resume to accomplish this. You can create an ATS-compatible resume, which allows recruiters to search your resume on any job board.
2. I recommend that you use one of the following resume-building websites, which are both elegant and simple to use. — ResumePuppy , Later Resume , Overleaf
3. When creating your resume, don’t limit it to a single page (as it is for data science interviews) — try to include more information on Outcomes and How you solved the specific problem.
4. You can include your contributions, side projects, additional coursework, or learnings alongside your professional projects.

Job Boards

1. Once you have your resume ready, Obviously you will post it on “Naukri” and you can even post it on “InstaHyre,” which is more efficient in terms of getting interview calls and conversions, as I personally believe that applying directly through the company websites does not yield any results.
2. Frequently updating your profile and having relevant skillsets listed will help you get more interview calls

Linkedin and Networking

1. Open to Work will also place your profile at the top of search results for recruiters on Linkedin.
2. It will also inform your network that you are looking for a job change, allowing them to assist you in obtaining referrals.

Learn — ReLearn — UnLearn

The most important portion of your job search is already underway. While the earlier steps have prepared you for it in certain ways, this stage requires greater focus in order to land better positions.

As I previously stated, a data science interview entails 4–6 rounds of interviews in addition to an HR talk. These 4–6 rounds can be divided into 4 different categories. Even with only three rounds, this is still possible, but the goals remain the same. Technical Assignments / MCQ / Coding rounds

1. Personal / Panel Interview on Technical Concepts
2. Business / Case study rounds
3. Leadership rounds.

Out of the above 4, 3 rounds are purely technical and the final one is a mix of Managerial, Behavioural + Culture fit round with a bit of alignment of roles and responsibilities (this is much more important in Startups ) — Tbh I got rejected in a couple of leadership rounds as I could not align on their expectations. To Crack the other three types of rounds, the Learn-ReLearn-UnLearn is the well-known secret mantra.

Learn New things to you: You need to master a few more concepts and algorithms to live up to expectations, according to your self-evaluation. You may not have utilized “Naive Bayes” or “agglomerative Clustering” in your projects, but as a DS, it is typical for the panel to ask questions about fundamental machine learning techniques, so you should still brush up on the basics. If it isn’t listed on your CV,

ReLearn your experience: Although you may have used numerous algorithms and completed numerous projects, your busy schedule may have prevented you from thoroughly studying those principles. List the algorithms that are mentioned in your resume before the interview and get ready for any reasoning inquiries that will revolve around those ideas. Relearn why something is named XGboost, for instance, if you have dealt with it before. How is missing value handled by XGBoost? When should we avoid using SMOTE? and so on

UnLearn Misconceptions: Always be sure to un-learn everything you may have understood incorrectly. For instance, numerous narratives have been written on overfitting, dealing with imbalanced datasets, employing dropouts, the need for deep learning for tabular data, and the idea that prophet is the most effective method for forecasting difficulties.

Present Yourself :

After completing your technical preparation, you should concentrate on honing your presenting abilities so you can succeed in face-to-face interviews. a few pointers that could help you

1. Explicitly describing your prior experience can help your recruiter choose how best to use you.
2. Giving more details on HOWS rather than WHATS would boost credibility and confidence.
3. Having a clear understanding of your goals and objectives might help you determine whether or not you’ll enjoy this position.
4. Instead of choosing jobs solely on the basis of CTC, consider the category’s future, growth, and learning opportunities.
   While responding to the questions, case studies, and problem-solving rounds, you should demonstrate your individuality without sacrificing the essentials.
5. For example, If you are required to develop an ML pipeline for an ETA estimation problem, for instance, demonstrate how structurally you approach the problem by listing and testing hypotheses, applying them in feature selection, and formulating it correctly so that you can build a specific ML Model.

This brings the Framework for planning your interview preparation process to a successful conclusion. Here I am including the skillset expectations from most of the job descriptions

I hope this article will assist you in taking the next step toward a fruitful job search.

If you thought this article was helpful, please let me know by sharing it with your network.

Follow me for more posts like this!

Comments are appreciated! Contact me via LinkedIn if you want to work together. #LearnWithKaran #PrepWithKaran

Happy Job Hunting! Happy New Year 2023!