A deployed XGBoost model on a SageMaker endpoint receives live inference requests via an AWS Lambda function. The team needs to detect if the model's accuracy degrades over time using incoming live data. Which solution fulfills this need?

The correct solution is to schedule a SageMaker Model Monitor job. SageMaker Model Monitor is specifically designed for detecting data and model quality issues, including drift, in deployed models. It works by comparing real-time inference data (inputs and/or predictions) against a baseline established from the training dataset's statistics and constraints. This allows for automated detection of significant deviations that could indicate model degradation. Building a CloudWatch dashboard would only visualize metrics, not automatically detect drift. Modifying the Lambda function to compute drift would be a complex and custom solution, duplicating functionality already provided by SageMaker Model Monitor, and would be less scalable and maintainable. SageMaker Debugger is used for monitoring training jobs to identify issues like vanishing gradients or overfitting, not for post-deployment model drift detection.