AI Engineer: MLOps and Model Deployment Flashcards
6 cards from real AI practice questions. Tap to flip, then mark Knew It or Still Learning — missed cards come back until you master them.
Read the first 6 AI Engineer: MLOps and Model Deployment flashcards as text
Which MLOps practice ensures that a model retrained on new data maintains or improves its performance compared to the previous version?
Answer: Model validation and comparison
Model validation and comparison ensures newly trained models meet or exceed baseline performance before being promoted to production.
What is the primary purpose of a feature store in an MLOps pipeline?
Answer: To provide a centralized repository for reusable, versioned feature data
A feature store centralizes feature computation and storage so teams can share, reuse, and version features consistently across training and serving.
Which deployment strategy releases a new model to a small subset of users before rolling out to everyone?
Answer: Canary deployment
Canary deployment directs a small percentage of traffic to the new model, limiting blast radius if issues arise.
What does 'data drift' mean in the context of production ML models?
Answer: The statistical distribution of input features changes over time
Data drift occurs when the real-world input distribution shifts away from the distribution the model was trained on, degrading performance.
Which container orchestration platform is most commonly used to deploy and scale ML model serving workloads?
Answer: Kubernetes
Kubernetes automates deployment, scaling, and management of containerized applications, making it the standard for production ML serving.
In CI/CD for ML, what does a 'model registry' primarily store?
Answer: Versioned trained models with metadata and lineage
A model registry stores versioned model artifacts along with metadata like metrics, lineage, and stage (staging, production) for governance.