What did you do in your previous job? What tools did you use? Did you process large datasets? Do you write code and how do you ensure clean code practices? What are your MLOps capabilities? What problems did your solutions have in your previous work and how did you fix them? They also asked classic ML questions: When is classical machine learning better than deep neural networks? What is boosting, bagging, XGBoost, random forest, and decision trees in general?
Could you please descrive the challenges you had during your recent projects and how you solved them?
Three take home coding challenges.
Design a CVR prediction model
Questions based on previous work experience. Scenario-based questions to gauge the knowledge of Big Data, System Design, etc. Questions on Machine Learning Training, Retraining pipelines, Automated Deployments, etc.
ML algorithm using numpy/pytorch. System design around video search system.
How would you handle tracking different colour balls in a cricket setting? (1 colour may be harder, and may have less data)
Compare advantages/disadvantages of CNNs vs Transformers
Basic data science questions for the most part, something like what is p-value, what is overfitting, cross-validation etc.
implement rate limiter. classify documents into two categories.
Viewing 571 - 580 interview questions