Interview Practice Questions

What probability distributions are commonly encountered in machine learning?

How can you optimize ML model performance?

What happens when there are two dominant classes in multiclass task?

When to use complex models versus simple analytics?

What key metrics do you use to evaluate ML model quality?

How do you envision key stages of ML project?

What can influence bias increase in model with unchanged variance?

What actions can lead to model variance increase?

How to formulate multiclass task via binary classification?

How to measure ROI (return on investment) from ML project?

How to organize monitoring and alerting system for ML models?

Where to find help when ML model issues arise?

How to properly conduct statistical analysis of ML experiment results?

How do recommendation systems and personalization work?

What methods exist for ML pipeline automation and notifications?

When to develop model from scratch instead of using existing solutions?

How to work with ML in limited infrastructure conditions?

If new products are added, do you retrain models here?

In which models is regularization a good idea?

What methods do you know to work with class imbalance?