How to give a good interview for AI/ML related jobs.

1. Why data science and why this company

All the interviews start with talking about yourself and why you are interested in data science. You need to talk about how what you are doing (University/Job) is related to data science. If what you are doing is not relevant to the job role, you may talk about what you have learnt about data science - Kaggle/personal projects/blogs/conferences etc.

2. Know about your previous projects in detail

For the projects you have mentioned in the resume, ensure that you know all the nitty-gritty of the project. Starting from the problem statement, data collection/exploration, model building and deployment. It’s a good idea to recollect and write down exactly what you will be talking about the project. For example, know why you used a particular ML model? How did you tune the model? What kind of recommendations did you make? What was the business impact of your project?

3. Learn about the company

Know well about the company you have applied to. Apart from the basic information such as revenue, different functions, the company motto and beliefs, find out more about the company’s clients or products. Learn about the projects they are working on and ask questions around it. Also, it is good to talk about how you can add value to the company in general.

4. Know SQL and Excel

Interviewers assume everyone knows Excel — and hence you won’t be tested on it. Along with Excel, SQL is a must-know for every data scientist. All the data querying will require SQL and you will definitely be tested on writing SQL queries. So you have to be quick on your feet as you may not be allowed to use any online help.

5. R/Python

Most of the R/Python interviews were given as a take-home test. The take-home test is very similar to a problem you would find in Kaggle. It is well defined and the data is fairly clean. However, you would be expected to be thorough in your analysis- starting from EDA to building multiple models and tuning your model, and finally providing insights and recommendations. The choice of language is yours — R or Python. Also, using Jupyter notebook for the test is recommended for better readability.

6. Basic Statistics

Go through the elementary statistics you would have learnt in college — Probability, Statistical Inference, Hypothesis Testing, Central Limit Theorem, Law of Large Numbers etc. A good understanding of p-value, confidence intervals and hypothesis testing is required. Also, practice probability questions on Conditional probability, Bayes theorem.

7. Machine Learning

It’s necessary to have a thorough understanding of some of the basic ML algorithms such as Linear Regression, Logistic Regression, kNN and k-means clustering. Questions on the basic algorithms can be rigorous such as assumptions of linear regression, how do you perform backward elimination, what are the different parameters you check to build a regression model etc. And, it is good to have a general idea of the other algorithms like Lasso/Ridge regression, SVM’s, Neural Network etc. Also, it is definitely okay to say you don’t know about a specific algorithm as long as you don’t have it in your resume.

8. Open-ended data science problems

In a lot of interview rounds, some were asked an open-ended data science problem and this is to check your thought process for solving a problem. My tip is to never jump to a solution directly, take your time, think out loud, and think of the various factors that affect the problem’s objective. As an example, it is asked how he/she will approach a dynamic pricing problem for a flight. So, it’s good to start with the drivers of flight fares— duration of the flight, time of booking, source, destination, fuel charges, demand etc. And then maybe you can talk about what algorithm/methodology you can use to combine the various factors. For example, multiple regression can be used to predict the optimum pricing of the flight. The interviewer won’t expect you to reach an optimal answer or any sort of conclusion- however, this is just to check your ideas and approach to solving a problem.

9. Data Infrastructure

This was rarely asked in interviews, and it was mostly asked in start-ups and smaller sized companies where there is a significant overlap between the roles of data scientists and data engineers. In such cases, it’s good to know Hadoop, Spark, cloud services such as AWS, Azure, and other technologies which they would have mentioned in their job role.

10. Project Management

There are a few questions regarding project management abilities. Specific questions involved with previous roles in projects, how it is worked with cross-functional teams, the challenges were faced, and experience leading a team etc. Some questions revolved around specific instances during the project where someone have walked the extra mile to ensure the success of the project.

These are the top areas around which most of the questions were asked. Only a few questions were asked about academics — because people have years of prior experience in Data Science. The most important aspect of the interview process is to learn after every interview. Many questions are repeated in interviews, so there is a high chance of getting them again.