What is assessed.
WHAT IS TYPICALLY COVERED
Fundamentals: evaluation, overfitting, algorithm behaviour
Coding, usually in Python Data manipulation and SQL System design for machine learning Discussion of your past work
WHAT FUNDAMENTALS QUESTIONS PROBE
Whether you understand what models do, not whether you can recite definitions.
WHAT COMMON QUESTIONS ARE
How would you detect overfitting When is accuracy the wrong metric How would you handle imbalanced classes How would you detect leakage Why might a model that tested well fail in production
WHAT SYSTEM DESIGN QUESTIONS ASSESS
Whether you think about data, deployment and monitoring, not only modelling.
WHAT TO COVER IN THOSE ANSWERS
Problem framing and metric choice Data sources and labels Baseline Evaluation strategy Serving pattern Monitoring and retraining
WHY THE BASELINE MATTERS IN AN INTERVIEW
Mentioning it signals practical experience immediately.
WHAT TO PREPARE ABOUT YOUR OWN WORK
What the problem was What you decided and why What went wrong What you would do differently
WHAT TO NEVER OVERSTATE
Results, or your role in them.
