Question 1

A manufacturing company uses a legacy on-premises ERP system to track assembly line issues. The company wants to integrate Amazon Bedrock to automatically analyze "Technician Notes" (unstructured text) and categorize the root cause of failures using the Anthropic Claude 4.5 Sonnet model.The legacy ERP system has a strict limitation: it blocks all user operations while waiting for external API responses and times out if a response takes longer than 300 milliseconds. The GenAI analysis typically takes 3–5 seconds to complete.As a GenAI Developer, which integration architecture should you design to incorporate the Foundation Model (FM) capabilities without disrupting the legacy system's operations?

Options :

A : Use AWS AppSync with a GraphQL subscription. configure the ERP system to subscribe to the RootCauseAnalysis mutation. When a note is saved, trigger the mutation to invoke Amazon Bedrock directly. The subscription connection keeps the channel open without blocking the main ERP thread, allowing the response to stream back when ready.

B : Create an Amazon API Gateway REST API with a synchronous integration to an AWS Lambda function. Configure the ERP to call this API with the notes. Update the Lambda function to increase the memory allocation to 10 GB, ensuring the model inference completes within the 300-millisecond window.

C : Implement an asynchronous, event-driven architecture. Configure the ERP to push the technician notes to an Amazon SQS queue and immediately return success to the user. Use an AWS Lambda function to poll the queue, invoke the Amazon Bedrock model for analysis, and write the classification results back to the ERP database via a separate callback API.

D : Deploy an AWS PrivateLink connection between the on-premises network and the Amazon Bedrock service endpoint. Configure the ERP system to make direct HTTPS calls to the Bedrock InvokeModel API using the private endpoint, bypassing the public internet to reduce network latency below the 300-millisecond threshold.

Answer: C

Question 2

A financial services firm processes thousands of unstructured PDF earnings reports daily. The engineering team uses Anthropic Claude 4.5 Sonnet on Amazon Bedrock to extract specific Key Performance Indicators (KPIs) and populate a downstream SQL database.Users have reported that while the model handles narrative text well, it consistently hallucinates values when processing financial tables that span multiple pages or contain complex nested headers. The raw text ingested from the OCR pipeline often arrives with broken formatting, missing whitespace, and merged columns.Which strategy should the GenAI Developer implement to enhance the input data quality and resolve the extraction errors?

Options :

A : Use an AWS Lambda function to apply a uniform truncation to all input documents, limiting them to the first 2,000 tokens to ensure the model focuses only on the high-level summary tables usually found at the beginning of reports.

B : Increase the temperature parameter of the Claude 4.5 Sonnet model to 0.7 to encourage the model to be more creative in interpreting the broken table structures.

C : Introduce a pre-processing step using the Amazon Bedrock ConverseAPI with a lightweight model like Anthropic Claude 3.5 Haiku. Prompt this model to rewrite the raw OCR output into valid Markdown tables, then pass this cleaned, structured text to Claude 4.5 Sonnet for the final KPI extraction.

D : Replace the OCR pipeline with Amazon Textract's raw text detection. Directly concatenate the raw LINE block outputs into a JSON object and feed this massive JSON structure into Claude 4.5 Sonnet without any further serialization.

Answer: C

Question 3

A healthcare technology company is building a patient inquiry chatbot using Amazon Bedrock. Before logging conversation transcripts for quality assurance, they must ensure that all personally identifiable information (names, phone numbers, addresses, medical record numbers) is detected and redacted to comply with HIPAA privacy requirements.Which combination of AWS services should the developer use to detect and redact PII from conversation logs?

Options :

A : Use AWS CloudTrail to automatically redact PII from all logs

B : Use Amazon Textract to extract PII from documents, then delete the documents

C : Use Amazon Comprehend to detect PII entities, then use Amazon Bedrock Guardrails to redact them from logs

D : Use Amazon Macie to scan conversation logs stored in S3

Answer: C

Question 4

A financial services company uses Amazon Bedrock to generate investment analysis reports. The prompt engineering team frequently updates prompts to improve output quality, but recent production incidents showed that prompt changes caused incorrect financial calculations.The company requires a solution that enables prompt versioning with the ability to rollback to previous versions, tracks which prompt version generated each report for audit compliance, sends alerts when prompt performance degrades (measured by downstream validation failures), and maintains a complete audit trail of all prompt changes for regulatory reviews.Which solution meets ALL these requirements with the LEAST operational overhead?

Options :

A : Use Bedrock Prompt Management to create and version prompts. Store each prompt version with metadata tags indicating the version number. Configure CloudWatch Logs to capture all Bedrock InvokeModel requests with prompt details. Create a Lambda function that parses CloudWatch Logs, detects validation failures, and publishes alerts to SNS. Store prompt change history in DynamoDB for audit reports.

B : Use Bedrock Prompt Management to create versioned prompts with immutable versions and aliases. Tag each report with the prompt version/alias used during generation and log this to CloudWatch Logs. Enable CloudTrail to capture all Bedrock API calls including prompt updates. Create CloudWatch metric filters on validation failure logs and configure CloudWatch Alarms to trigger SNS notifications when failure rates exceed thresholds. Use CloudTrail Event History and CloudWatch Logs Insights for audit trail queries.

C : Store prompts as text files in S3 with version control enabled. Implement a Lambda-based prompt versioning system that reads prompts from S3, tracks versions in DynamoDB, and passes prompts to Bedrock InvokeModel API. Configure S3 Event Notifications to trigger Lambda functions that log prompt changes to CloudWatch. Use CloudWatch Synthetics to test prompt quality and send alerts via SNS when tests fail.

D : Use AWS Systems Manager Parameter Store to manage prompts with versioning enabled. Create a Step Functions workflow that retrieves the prompt version from Parameter Store, invokes Bedrock, and logs the prompt version and output to CloudWatch Logs. Enable Parameter Store change notifications via EventBridge to track prompt updates. Create CloudWatch Alarms based on custom metrics published by the Step Functions workflow when validation failures occur.

Answer: B

Question 5

A bank uses an AWS CodePipeline to deploy a GenAI loan approval application. The application consists of an Amazon Bedrock Knowledge Base and a set of Lambda functions. The compliance team requires that before any new Knowledge Base configuration (e.g., chunking strategy change) is promoted to Production, it must pass a "Golden Set" evaluation.If the evaluation score (F1 score) is below 0.85, the pipeline must automatically rollback.Which solution meets these requirements?

Options :

A : Use Amazon Inspector within the pipeline to scan the Knowledge Base configuration for misconfigurations and vulnerabilities.

B : Add a Manual Approval action to the pipeline. Configure an Amazon SNS notification to the compliance team with a link to the Bedrock Console so they can manually test the agent and click "Approve."

C : Add a Test Stage to the pipeline. Use AWS CodeBuild to run a script that invokes the new Knowledge Base Retrieve API against the golden questions. Calculate the score and return a non-zero exit code if it fails, causing the pipeline to stop.

D : Use AWS CloudFormation Guard. Define a policy that checks the VectorIndexConfiguration parameter in the template. Fail the stack deployment if the chunking strategy has changed.

Answer: C