En route to Microsoft Data Scientist

In this post, I'll be talking about the selection process of Microsoft for data scientist profile in IIT Bombay.

Microsoft visited our campus for recruitment in two profiles:

1.  Software Developer
2. Data Scientist

I will mostly be talking about the selection process of the data scientist profile.

First Round - Online Coding Test and Data science MCQs

For Software Developer profile, we had three coding questions to solve on cocubes platform. All the students were having a different set of questions.

1.  Find the distance between two nodes of a binary tree
2.  Given a string s, find the next permutation of s
3. Parse a string s and convert numbers into their corresponding characters, remove the whitespaces and number followed by hash # is not converted into character.
• Sample Input -  3 2 20_21 # 1 # 2_#11 4 @11
  Sample Output - CBT U12 11D@H

For Data Science Profile, there were a lot of MCQs on general Machine Learning concepts like AUC-ROC curve,L1/L2 Regularization Random Forests, Bagging, Bias and Variance. Some of the MCQs were on advanced ML concepts like Convolutional Neural Nets, Recurrent Neural Nets, Dropouts and Batch Regularization.

Everybody was allowed to give both the tests. A preference list for profiles is taken for each student.

This was an elimination round. After the first round, two different shortlists for two different profiles were out. If a person did well in both tests, he was put in shortlist of his preferred role. A candidate can only be in shortlist of any one profile.

I got selected in data science profile.

Second Round - Group Fly (Pen and Paper Mode)

For Software profile, there were two coding questions for which you have to write proper code on paper in some language like C/C++/java (not pseudocode).

1.  Run-length encoding and decoding
2. Another question on binary tree

For Data Scientist profile, there were again two questions based on ML concepts. We have to answer using pen and paper.

1.  Basic ML: Write mathematical derivations of one of these:
• SVM
• Neural Networks
• Hidden Markov Models
• Conditional Random Fields
2. Advanced ML: A real-world problem is given to solve. Write data collection, preprocessing and feature extraction, a model for prediction, mathematical description of the model if possible, alternative approaches for solving the problem. Solve any one problem.
• E-commerce recommendation system
• Given a search query, how likely a user will click a particular link

Shortlists for each profile for interviews was out afterwards. Almost half of the crowd was eliminated.

Third Round - Interviews

I will be telling the interview experience for data scientist profile.

1.  First Interview round: This was a pure technical round.
• He asked to briefly describe all the projects in my resume. All of my projects were mostly related to image processing and machine learning.
• Then, he asked me to design a movie recommendation system. We had a quite good discussion over that where I was speaking for 80% of the time.
• He asked a coding question: nth-node from linked list. I saw this question for the first time. So, I started with brute force and finally reached the optimized solution using two pointers.
• In the end, we had a discussion about work going in the ML field in India.
2. Second Interview Round: This was a pure technical round too.
• I was asked to describe my M.Tech Project in detail.
• He gave me a couple of real-world problems related to machine learning and asked me how will I approach it.
• One problem was if a satellite has taken an image which is occluded by clouds and we want to recover the image without any occlusion. We are also given a drone which can click images. How can we use drone images to recover our satellite image?
•  Another problem was if a search query is given, how will you convert it into a question? For e.g: Input: sky blue reason, Output:  why is sky blue?
• In the end, we had a discussion about work going in the image processing with ML field in India.

Both the interviewers seemed happy with me. That's why there were no further technical/HR rounds. HR came and said Congratulations to me. She mad me meet Asia HR head and General Manager of AI and Research - Mr Sundar Srinivasan.

In the end, they announced the selected students and gave them goodies. This is how I received an offer of data scientist at Microsoft :D.

Let me know in the comments section if you have any doubt about the experience.