InfoCusp Interview Experience for Machine Learning Engineer

Applied off-campus for the post of Machine learning Engineer. There are a total of 8 rounds. 4 ML rounds, 2 software rounds, 1 HR, and 1 coding.

Rounds 1 and 2 were of ML, 3rd was to test my software skills, 4th and 5th were again ML, 6th was HR, 7th was SW one and final coding round.

Questions asked in ML rounds :

1. Tell me the real use case where you would use precision and recall.
2. What are the different types of loss functions. When you will use MSE and MAE. Write the formula for binary cross entropy.
3. How do you know your model has overfitted? How to overcome overfitting.
4. One question was on Bayes theorem didn’t remember properly.
5. If you have a bunch of resumes and you want to let say group resumes for a company, how will you do that ? So basically it was related to clustering, and we can use embeddings like word2vec for skills and important field in resume and use some distance metric to cluster. We can use cosine similarity to find the nearest matching resume also.
6. I had used BERT in one of my project so asked about word2Vec embedding and Bert embedding. What is the difference.
7. If LSTM and BERT had same parameters than which would be faster to train. So basically she was looking for parallelism thing like in LSTM we need to give input one at a time while in Bert we give at once.
8. Different techniques to do over sampling of minority class for images and in general.
9. What are hidden Markov models.
10. Told me to design XOR using MLP.
11. Will you normalize data at the time of testing, yes or no? If yes than how.
12. Given a paragraph, how will you give a title to paragraph if suppose the answer lies in the paragraph itself. Asked me how the dataset would look like.
13. Explain Bert in detail.
14. Explain Bagging and Boosting and their difference.
15. Assumptions of Naive Bayes.
16. Why do we use activation functions ?
17. Maximum Likelihood Estimation (MLE) vs Maximum A Posteriori (MAP) 
18. Asked what are anonymous functions in python and different ML Libraries I know.
19. What are eigen vectors and eigen values. So just go through basics of it and also SVD and covariance and all.
20. Asked different dimensionality reduction technique I know so i explained PCA.
21. How to handle imbalanced dataset – he was particularly looking for class weight method.

Questions asked in Software rounds:

1. Given 4 points, how will u determine if they form a square or not?
2. Given an array of integers where each integer lies in the range 0 to n-1, find first repeated number. 
3. Given array of n integers where each integer lies in range 1 to n, one now is missing and one is repeated. Find both of them.
4. Design Among us game by writing classes and functions.
5. Difference between multithreading and multiprocessing. Why multithreading is used if we can achieve the same using multiprocessing.
6. How can we assign two IP address to a computer ?
7. Explain ACID properties, Normal forms, deadlock avoidance.
8. Delete the node, given pointer to that node.
9. Have a look on how to map relationship(one to many and all) into classes and concept of aggregation, generalization.
10. Explain OOPs concept and specially polymorphism.

Coding round :

1. A minimum number of dice throws to reach target in snakes and ladder game (10*10 board).
2. Given A, B, C, D, K and f(t)=At³  + Bt²  + Ct + D, find max t such that f(t)<=K.
3. Given n-array tree and number M, u can change nodes value between 1 to M in such a way that GCD of all root to leaf path is 1. Find the total number of such combinations.

HR round :

1. Why you chose CSE.
2. Family background.
3. Why would we choose you rather than any other candidate?
4. What is your definition of an idle organization?

Thank you for reading !