MLA-C02 mock exam

Question 1, Data Preparation for ML and AIYou must measure whether a training dataset is imbalanced across a sensitive attribute before training. Which tool fits?

• AAmazon SageMaker Feature Store
• BAmazon SageMaker Ground Truth
• CAmazon SageMaker Clarify
• DAmazon SageMaker Autopilot

Question 2, Data Preparation for ML and AIYou are preparing long PDFs for a RAG application. What step comes before generating embeddings?

• AModel distillation on the embedding model
• BChunking the documents and extracting metadata
• CConfiguring auto scaling for the vector store
• DHyperparameter tuning of the embedding model

Question 3, ML Model and Foundation Model (FM) DevelopmentA model works well in training but fails on validation data. Which change most directly reduces overfitting?

• AAdd regularization or simplify the model
• BTrain for more epochs until training loss stops falling
• CAdd more features from the raw data
• DRemove the validation set and train on everything

Question 4, ML Model and Foundation Model (FM) DevelopmentYou need to compare the quality of generated summaries against reference summaries. Which metric is meant for this?

• AR-squared
• BRMSE
• CROUGE
• DAUC

Question 5, Deployment and Orchestration of ML and AI WorkflowsTraffic is sporadic with long idle periods, and a few seconds of cold start is acceptable. Which option is most cost-effective?

• AA batch transform job each minute
• BA dedicated GPU cluster
• CA real-time endpoint on large instances
• DSageMaker Serverless Inference

Question 6, Deployment and Orchestration of ML and AI WorkflowsA job takes 30 minutes per request on large payloads, and clients can pick up results later. Which option fits?

• AA real-time endpoint
• BAsynchronous inference
• CAn EC2 cron job
• DServerless inference

Question 7, Operating, Monitoring, and Securing ML and AI SolutionsModel quality slowly degrades in production with no change to code. What should you set up?

• AAmazon SageMaker Model Monitor with drift alerts
• BA bigger instance type with more memory for the endpoint
• CA new VPC with stricter routing for inference traffic
• DA larger S3 bucket to retain more inference logs

Question 8, Operating, Monitoring, and Securing ML and AI SolutionsAn endpoint should only be reachable from your VPC, with least-privilege access. What do you configure?

• AAn S3 bucket policy that denies public access
• BA public endpoint with a shared API key in Secrets Manager
• CA VPC interface endpoint and a tightly scoped IAM role
• DA wider security group rule limited to your office range