ZAPR Media Labs Research Scientist Interview Questions | Glassdoor.co.in

ZAPR Media Labs Research Scientist Interview Questions

Interviews at ZAPR Media Labs

2 Interview Reviews

Experience

Experience
100%
0%
0%

Getting an Interview

Getting an Interview
50%
50%

Difficulty

3.0
Average

Difficulty

Hard
Average
Easy

 

Research Scientist Interview

Anonymous Employee in Bengaluru
Accepted Offer
Positive Experience
Easy Interview

Application

I applied in-person. The process took 2 weeks. I interviewed at ZAPR Media Labs (Bengaluru) in October 2019.

Interview

The whole process was made smooth and easy by both technical and HR team. Overall the Interview process consist of 4 rounds.

Round 1 - Introduction (brief conversation to get an idea of my interest and current area of work going in Zapr)
Round 2 - Technical (Key topics - Basics of Machine learning and Speech processing and my current work)
Round 3 - Meet the founders (Face to Face interview with the ZAPR's CTO (Sajo Mathews)
Round 4 - HR round

Interview Questions

  • How regularization is done in the Neural network? How does weight change after applying it?   Answer Question
  • In stock market prediction, what kind of ML algorithm can be used, type of loss functions, and type of activation functions are used?   Answer Question
  • Is it possible to do the PCA operation with Neural network? How does an Autoencoder work? What kind of loss functions are used?   1 Answer
  • Difference between Wasserstein loss vs L2 loss in the context of GAN?   Answer Question
  • How with limited data or single reference sentence prosody transfer can be done in Tacotron or any other network?   Answer Question

Other Interview Reviews for ZAPR Media Labs

  1.  

    Research Scientist Interview

    Anonymous Employee in Bengaluru
    Accepted Offer
    Positive Experience
    Difficult Interview

    Application

    I applied through an employee referral. The process took 2 weeks. I interviewed at ZAPR Media Labs (Bengaluru) in December 2018.

    Interview

    Telephonic discussion was mostly technical with focus on machine learning and data preprocessing. I will focus mostly on the on-site interview as it reflects the experience better.

    On-site interview process is completely hands-on. I was given a dataset with a few days time limit to explore in my own time. In the interview I was assessed based not just on my exploration of the data but also on the improvements I can make on the models during the interview. The process is rewarding because you are constantly guided by the interviewer and the results are awesome.
    This is followed by an in-depth discussion on architecting a pipeline/data model for a hypothetical company, which can be from any domain. For example I was asked to build a taxi app from scratch discussing all the elements that go into the system, from data to algorithms. This process is again a discussion rather than a QnA session.

    Interview Questions

    • All questions in the on-site interview were based on possible improvement to the models that I had built. and were from these topics.
      1. Dimensionality reduction (followed by using it)
      2. Train multiple models on the dataset judiciously
      3. Selecting the ideal metric for the dataset
      4. From scratch ensemble learning on the dataset with pre-prepared models
      5. Architectural questions and discussions on taxi service
      6. How will you use LSTM for analysing time-series data
      7. Time-series analysis to find outliers and trends   1 Answer

Don't Miss Out On a Job You Love
Upload a CV to easily apply to jobs from anywhere. It's simple to set up.