ML & Data Science · INTERVIEW & OA PREP
Machine Learning Fundamentals
ML and data-science interviews test whether you understand the concepts, not whether you memorised library calls. Bias-variance, overfitting, evaluation metrics, and how common models work are the recurring themes.
Key concepts
- Supervised vs unsupervised vs reinforcement learning.
- Bias-variance trade-off; overfitting vs underfitting and how to detect each.
- Train/validation/test split and cross-validation — why they exist.
- Metrics: accuracy vs precision/recall/F1, and when accuracy misleads.
Questions that get asked
- Explain the bias-variance trade-off.
- How do you handle an imbalanced dataset?
- Precision vs recall — when do you optimize for each?
- How does a decision tree decide splits?
- What is regularization and why use L1 vs L2?
Tip
Tie every concept to a concrete scenario ('for fraud detection, recall matters more because...') — applied reasoning scores highest.
Common mistake
Quoting accuracy on imbalanced data. A 99%-accurate model can be useless if 99% of samples are one class.
Quick checks
Three tap-only questions to test your recall on this topic.
Keep going
Ready to apply? Browse live internships and sharpen your resume first.