Siemens Deep Learning Engineer Interview

Ths Siemens Deep Learning Interview has the following stages:

• Resume Shortlisting by the Hiring Manager
• Initial Telephonic Round
• Onsite interview - whole day event

The Hiring Manager shortlists the resumes that he finds suitable for the position depending upon several criteria, like, 

• Educational Background
• Number of years of experience
• Number of projects
• Number of course work - either school or independent online courses
• Number of papers published
• Number of patents filed
• Depending on the role

• ML Researcher - usually a Ph.D. is preferred as the candidate shows more commitment /patience/maturity to the subject.  
• ML developer can be any software engineer with working knowledge of the ML.

Initial Telephonic round:

This is going to be a general interview on ML. The basic idea here is to judge the knowledge base of the candidate and make a decision if he is worth bringing for a Face-to-Face discussion onsite.

The usual questions asked in this telephonic round can be any ML-related topics. For more questions please read my other articles here: ML questions and DL questions.

Onsite Interview:

Depending upon the role the candidate can be asked to do a project presentation or a will directly move to the other round based on the hiring team.

The first round: Project Presentation

This can be any project you have done in the field of Machine Learning/ Deep Learning that you feel is good to present to the hiring team. The audience is going to be your future team members :) So do the best you can to make them feel that you are going to be a valuable resource person.

Next Rounding - Coding a DL/ML project:

The team decides a project that you need to do given a time considerations of 1-2 hours. The project can be anything the team thinks is a must know for the candidate. You can use any tools/library that you feel is needed to get the work done.

Sometimes the team also gives a laptop without any setup done. The candidate must get the setup done and then start coding the DL project. The setup can be right from selecting the Cuda libraries to the framework - tensorflow / pytorch installation to the coding IDE of your choice.

Once the code is completed there will be a code review round. Here the candidate has to present his code and suggest possible improvements to it. 

The Next round:

If you clear code round, you will be next interviewed for the technical concepts - usually theory and mathematical proofs of the ML/DL concepts. 

Initial questions will be in the ML domain, most of the questions can be read here ML questions Then questions on Deep Learning starts.

If you are more into perception domain then it will be more into the CNN architecture and if its NLP (more into sequence models) then more in the RNN domain.

The idea is to test the basic understanding of the candidate in the ML/DL field so that he can learn as new architecture are brought in the field.

Manager Round:

This the round with the manager of the team. Usually a low technical and more behavioral round. The basic idea is to gauge the candidates profile both EQ and IQ.

HR round:

If you are found to be the suitable candidate next would be the HR round. This can be done either on the same day or sometime later on the telephone to discuss the pay package and joining dates. 

All the best for your interview.

Do let us know how your interview experience was and any questions that were interesting to be added in this list. 

As I always say Deep Learning => Keep Learning :)

Most Common Interview questions for a Machine Learning Engineer - PART 2/2

These are some of the most common interview questions for Machine Learning Engineer

PART 2/2

1. How do we address High Bias and Variance scenarios?
2. What is regularization?
3. What happens if we reduce the number of features - what is affected bias or variance or both? Explain
4. What is L1 Regularisation?
5. What is L2 Regularisation?
6. Which is sparse - L1 or L2 regularisation?
7. What are Neural Networks?
8. What is Forward prop and Backward prop?
9. What is Gradient Checking?
10. How do we do weights initialization?
11. Explain a typical ML pipeline?
12. How do we evaluate a Learning algorithm?
13. How do we do the model selection?
14. What is Linear Regression with Regularisation?
15. Plot a graph of error Vs the number of training examples? The error is Cross-validation and Training losses.
16. Will collecting more data solve - high variance or bias?
17. What is Precision / Recall / F1 score ?
18. What is SVM?
19. Why is SVM called Large Margin Classifier?
20. What are the kernels in SVM? Name some kernels.
21. When to use Logistic Regression and SVM?
22. Is SVM a convex or concave function?
23. Does SVM have a global optimum?
24. What is unsupervised learning?
25. What are the types of unsupervised learning algorithms?
26. What is the K-Mean algorithm?
27. What is PCA?
28. Why is PCA used?
29. What is t-SNE?
30. What is the meaning of 99% of variance is retained mean in PCA?
31. What is Anomaly detection?
32. What is a Gaussian Distribution?
33. What is the difference between a Gaussian and Normal distribution?
34. What is Content-Based Recommendation?
35. What is Collaborative filtering?
36. What is Stochastic Gradient Descent?
37. What is Batch Gradient Descent?
38. What is Mini-Batch Gradient Descent?
39. What is Map-reduce?
40. Can you explain any ML pipeline with an example use case?
41. How do we do OCR?
42. What is a Decision Tree? Answer
43. What is a Random Forest? Answer

The Best Books recommended to improve your Machine Learning skills:

   

For Part 1 of the series click here: PART 1

For Deep Learning Interview Questions Please Check this series: Deep Learning 

KPIT Interview Questions

The selection procedure for KPIT job is done in the following steps:

1. Written test
2. Technical interview
3. HR round

Written test

• Few aptitude questions
• Technical questions on microcontroller, digital electronics, analog electronic - opams, CMOS,etc.
• C language questions

Technical Round Questions

1. Draw full wave rectifier circuit. Which is better bridge rectifier or centre tapped? 
2. How to verify the health of the 1kb RAM without using checksum.
3. Write a code for moving average when the input is analog data.
4. Draw a memory map for accessing external ROM and write a code to access the external ROM.
5. Design a mod3 counter asynchronous.
6. Transfer characteristics of a clipper was given and told to design the circuit.
7. Draw the output waveform for a RC (also RL) circuit with a step input taking the outputs across R and C separately.
8. More emphasis on projects :
• If you have used a microcontroller you should justify why you have chosen that microcontroller and not any other.
• How do you interface IR sensor to a microcontroller.
• Block diagrams and in depth understanding of each blocks.

This is the best book to learn Digital Design using Verilog HDL, VHDL and System Verilog

I suggest going through these books for Digital Electronics preparation:

>

HR round: (To read more about HR questions in other tech companies read here)

1. Tell me about yourself and your family.
2. Why should KPIT hire you? (Tell 3 strengths with suitable examples to substantiate the same)
3. Why do you want to work for KPIT?
4. Do you have any plans for further studies?
5. What are your achievements in college so far?
6. Do you have any specific questions regarding the job?

This is a book that I highly recommend for anyone to improve on their soft skills. This will not just help in your interviews but also in your day-to-day work.

Most Common Interview questions for a Machine Learning Engineer - PART 1/2

These are some of the most common interview questions for Machine Learning Engineer
PART 1/2

1. What are the types of Learning Algorithms?
2. What is Supervised Learning? And name its basic types?
3. What is Unsupervised Learning? And name its basic types?
4. What is a Linear Regression?
5. What is Logistic Regression?
6. What is a Cost Function?
7. What is a Convex Function?
8. What is Local Optima?
9. What is a Squared Error function?
10. What is the Cost function used for a Regression problem?
11. What is the Cost function used for a Classification Problem?
12. What is Gradient Descent?
13. What is the Learning Rate?
14. Is it necessary to change the learning rate during the training period?
15. What is Multi-Variate Linear Regression?
16. What is the difference between a Linear and Logistic Regression algorithm?
17. What is Feature Scaling?
18. What is Mean Normalization?
19. Is it better to have a small or large learning rate?
20. What is Feature Engineering?
21. What is Polynomial Regression?
22. What are the activation functions used in Linear Regression?
23. What are the activation functions used in Logistic Regression?
24. What are the different Optimization Algorithms?
25. Explain Gradient Descent, Conjugate gradient, BFGS, L-BFGS. What are the advantages and disadvantages of it? (This is a tough question)
26. What is multi-class classification and how do we do it ?
27. What is overfitting and underfitting ?
28. How do we address overfitting and underfitting?
29. What is High Bias?
30. What is High Variance?

The Best Books recommended to improve your Machine Learning skills:

   

I would also suggest having a look at the Udacity course Become a machine learning engineer.

For Part 2 of the series click here: PART 2