Top 10 Data Science & ML Interview Questions
Prepare for data science and machine learning interviews at European companies. Questions from companies like DeepMind, Spotify, Booking.com, and research labs across the EU.
Explain the bias-variance tradeoff.
Expert Tip: Use visual diagrams. Relate to overfitting/underfitting.
How would you handle class imbalance in a binary classification problem?
Expert Tip: SMOTE, undersampling, class weights, focal loss, evaluation metrics (F1, AUC-PR).
Walk me through building an end-to-end recommendation system.
Expert Tip: Cover collaborative filtering, content-based, hybrid approach, and A/B testing.
What is the difference between L1 and L2 regularization?
Expert Tip: L1 = feature selection (sparse), L2 = shrinkage. Geometric interpretation helps.
Design an A/B testing framework for a European e-commerce platform.
Expert Tip: Address GDPR data requirements, sample size, statistical significance, and multi-armed bandits.
Explain cross-validation. When would you use stratified k-fold?
Expert Tip: When class distribution is imbalanced — preserves ratio in each fold.
How does a transformer architecture work?
Expert Tip: Self-attention, multi-head attention, positional encoding, layer normalization.
What metrics would you use to evaluate a clustering algorithm?
Expert Tip: Silhouette score, Davies-Bouldin index, elbow method for k-means.
How would you detect and handle data drift in production ML models?
Expert Tip: KS test, PSI, monitoring dashboards, automated retraining pipelines.
Explain the difference between a generative and discriminative model.
Expert Tip: Generative: P(x,y), models joint distribution. Discriminative: P(y|x), models boundary.
Recommended Resources
Hands-On Machine Learning (Aurélien Géron)
Comprehensive ML book with practical Scikit-learn and TensorFlow examples.
Designing Machine Learning Systems (Chip Huyen)
Covers production ML — very relevant for European scale-ups.
Kaggle
Join competitions to build portfolio — European employers value this.