AIP-C01 AWS Certified Generative AI Developer – Professional Free Practice Questions and Answers

A team compares two Amazon Bedrock prompts and must monitor recommendation fairness across demographic groups, alert when a disparity exceeds a threshold, and produce recurring reports. Which managed service can evaluate bias metrics?

• A. Amazon SageMaker Clarify with metrics published to CloudWatch
• B. AWS CloudTrail Lake with a query for every response
• C. Amazon Route 53 health checks for each prompt
• D. AWS Budgets with a monthly cost threshold

A financial assistant must block specific high-risk advice topics, prevent competitor names in responses, and keep answers grounded in approved guidance. Which guardrail configuration best addresses these controls?

• A. Denied topics, blocked custom word filters, and a high grounding threshold
• B. A low grounding threshold and no output filters
• C. A latency alarm on the Bedrock endpoint only
• D. A prompt that asks the model to avoid all financial subjects without guardrails

A React application calls a Bedrock Knowledge Base through an AppSync GraphQL API. Long responses time out, especially for complex questions. The team wants to show text as it is generated with minimal custom work. What should it use?

• A. AWS Amplify AI Kit streaming responses
• B. A longer Lambda timeout with no client changes
• C. An SQS queue that the browser polls continuously
• D. A larger DynamoDB table for conversation history

A service must route Bedrock requests among foundation models using frequently changing cost, region-compliance, and user-tier rules. The rules should take effect without a code deployment. What is a suitable design?

• A. Store centrally managed rules in AWS AppConfig and have Lambda apply them per request
• B. Hard-code model IDs in each client application
• C. Update Lambda environment variables manually for every price change
• D. Create a separate API endpoint and deploy new code for each model switch

Several business units need consistent response structures but different tone and reusable prompt variants. The assistant must also block harmful content and sensitive terms. Which managed combination reduces maintenance?

• A. Bedrock Prompt Management with reusable variants and Bedrock Guardrails
• B. DynamoDB prompt records and custom post-processing for every response
• C. One Lambda function per department with no shared templates
• D. CloudWatch dashboards and an S3 bucket of prompt files