AIF-C01 Machine Learning Operations (MLOps) 2

Question 1

What is the key difference between a shadow deployment and a canary deployment?

Accepted Answer

Shadow deployment sends real traffic to the new model but discards its predictions

Answer

Shadow deployment routes all traffic to the new model without user awareness

Answer

Canary deployment tests only synthetic data while shadow uses real data

Answer

There is no meaningful difference between the two strategies

Question 2

Amazon SageMaker Feature Store serves what primary purpose in an MLOps workflow?

Accepted Answer

Providing a centralized repository for storing, sharing, and reusing ML features

Answer

Storing pre-trained model weights for foundation models

Answer

Managing hyperparameter configurations across training runs

Answer

Archiving historical model inference logs

Question 3

What is concept drift in the context of ML model monitoring?

Accepted Answer

A change in the underlying relationship between input features and the target variable

Answer

The gradual decrease in model accuracy due to infrastructure issues

Answer

When the input feature distribution shifts away from the training distribution

Answer

When a model is retrained with outdated data

Question 4

In Amazon SageMaker, what is a 'production variant' used for?

Accepted Answer

Splitting traffic between multiple model versions on a single endpoint

Answer

Defining different IAM permission sets for model endpoints

Answer

Specifying different training configurations for the same algorithm

Answer

Creating duplicate backups of a model artifact in S3

Question 5

Which SageMaker capability automatically detects bias in training data and model predictions?

Accepted Answer

SageMaker Clarify

Answer

SageMaker Debugger

Answer

SageMaker Data Wrangler

Answer

SageMaker Experiments

Question 6

What triggers are commonly used to initiate automatic model retraining in an MLOps pipeline?

Accepted Answer

New data availability, performance degradation, or scheduled time intervals

Answer

Only manual approvals from a data scientist

Answer

Changes to the AWS region or availability zone configuration

Answer

Updates to the SageMaker SDK version

Question 7

What is the role of SageMaker Experiments in an MLOps workflow?

Accepted Answer

Tracking, organizing, and comparing multiple ML training runs and their metrics

Answer

Running A/B tests on live production endpoints

Answer

Automatically deploying the best-performing model to production

Answer

Managing user access controls for SageMaker notebooks