Interview prep
Machine Learning Engineer interview questions
ML engineer interviews combine deep learning fundamentals, system design (serving, retraining), and production-scale coding. Expect questions on model architectures, optimization, and end-to-end ML pipelines.
🎤 Practice these out loud with AceCoach →The questions
Design an ML pipeline to rank search results in real time for a marketplace with 100M daily queries. System design
How to answerScope a feature store (offline compute, online serve), fast retrieval (top-K), ranking within a latency budget, and a feedback loop. Discuss latency vs accuracy and cold-start.
A model performs well offline (AUC 0.92) but 0.2 lower in production after a week. Debug this. Scenario
How to answerCheck data drift, feature-production bugs (encoding, missing values), and train/serve skew (batch vs online inference).
Implement the attention weights in a Transformer attention head. Coding
How to answerSoftmax(Q·Kᵀ / √d_k)·V; handle numerical stability (subtract max before softmax), validate output shape.
Tell me about a model you deployed that failed and how you recovered. Behavioral
How to answerSTAR: model + context, failure symptoms, root-cause and rollback decision, recovery time and process fix.
How would you cut CNN training time 50% without losing more than 2% accuracy? Fundamentals
How to answerMixed precision, gradient checkpointing, distributed training (DDP), LR scheduling, efficient architectures (depthwise-separable convs).
Design a feature store for real-time recommendations. How do you keep train/serve features consistent? System design
How to answerOffline batch compute, low-latency online cache, versioning, train/serve-skew detection, schema validation on ingest.
What machine learning engineers are tested on
PythonTensorFlow/PyTorchDistributed SystemsModel ServingFeature EngineeringMLOpsSQLDocker/Kubernetes
Frameworks are guidance, not scripts — the point is to make the answers your own. All roles →