AI AI Engineer: MLOps and Model Deployment 2

Question 1

What is the role of an A/B test in an MLOps deployment workflow?

Accepted Answer

To statistically compare two model versions using live traffic

Answer

To compress model size

Answer

To retrain a model on new data

Answer

To validate training data quality

Question 2

Which metric is most useful for detecting concept drift in a classification model?

Accepted Answer

Prediction distribution shift over time

Answer

Training loss

Answer

Inference latency

Answer

GPU utilization

Question 3

What is 'shadow mode' deployment in MLOps?

Accepted Answer

Sending production traffic to a new model without using its outputs to serve users

Answer

Running the model only at night to save costs

Answer

Encrypting model weights before deployment

Answer

Deploying a model without logging

Question 4

Which tool is commonly used for experiment tracking in ML, allowing teams to log parameters, metrics, and artifacts?

Accepted Answer

MLflow

Answer

Jenkins

Answer

Terraform

Answer

Prometheus

Question 5

What is 'online serving' in ML deployment as opposed to 'batch inference'?

Accepted Answer

Generating predictions in real time for individual requests

Answer

Training models using online learning algorithms

Answer

Serving models over the internet

Answer

Running inference on streaming data lakes

Question 6

Which practice helps prevent training-serving skew in production ML systems?

Accepted Answer

Sharing the same feature pipeline code between training and serving

Answer

Using different programming languages for training and serving

Answer

Retraining models weekly

Answer

Normalizing outputs post-inference