Loading...
Loading...
Technical questions for ML/AI engineering roles covering model development, MLOps, and statistical fundamentals.
7 questions with expert tips
Practice with AI →Explain the bias-variance tradeoff. How do you diagnose whether a model is suffering from high bias vs high variance?
Expert Tip
High bias: underfitting, high training error. High variance: overfitting, large gap between train and validation error.
Walk through how you would handle class imbalance in a fraud detection dataset where 0.1% are fraudulent.
Expert Tip
SMOTE/oversampling, class weights, threshold tuning, use precision-recall AUC instead of accuracy, cost-sensitive learning.
How do you detect and respond to model drift in production?
Expert Tip
Monitor input feature distributions, output distributions, and performance metrics. PSI for input drift, concept drift monitoring.
Explain the attention mechanism in transformers. How does self-attention allow models to handle long-range dependencies?
Expert Tip
Q, K, V matrices. Attention score = softmax(QK^T / sqrt(d_k))V. Each token attends to all positions equally.
You have a timestamp feature in a sales dataset. What derived features would you extract to improve a forecasting model?
Expert Tip
Hour of day, day of week, month, quarter, is_weekend, days_since_holiday, lag features, rolling averages.
A stakeholder insists the model needs 99% accuracy. It's a medical diagnosis system. What's your response?
Expert Tip
Reframe with precision/recall/F1 for imbalanced classes. For medical: sensitivity (recall) is critical—missing true positives is costly.
How would you design an A/B testing framework for ML models in production?
Expert Tip
Shadow mode, canary deployment (5% traffic), statistical significance testing, holdout sets, multi-armed bandit.
Answer these questions in a simulated interview and get real-time AI feedback on your responses, pacing, and depth.
Start Practice Interview