AIP Question Bank · Latest 50 Questions Free Preview (Page 2/5)

Questions 11–20 of the latest 50 AIP questions (5 pages total), with answers and explanations, no login required. Practice all 98 questions online.

Question 11 · No. 58 · Single choice

A company is building a generative AI (GenAI) application that processes financial reports and provides summaries for analysts. The application must run two compute environments. In one environment, AWS Lambda function must use the Python SDK to analyze reports on demand. In the second environment, Amazon EKS containers must use the JavaScript SDK to batch process multiple reports on a schedule. The application must maintain conversational context throughout multi-tum interactions, use the same foundation model (FM) across environments, and ensure consistent authentication. Which solution will meet these requirements?

A. Use the Amazon Bedrock InvokeModel API with a separate authentication method for each environment. Store conversation states in Amazon DynamoDB. Use custom I/O formatting logic for each programming language.
B. Use the Amazon Bedrock Converse API directly in both environments with a common authentication mechanism that uses IAM roles. Store conversation states in Amazon ElastiCache. Creating programming language-specific wrappers for model parameters.
C. Create a centralized Amazon API Gateway REST API endpoint that handles all model interactions by using the InvokeModel API. Store interaction history in application process memory in each Lambda function or EKS container. Use environment variables to configure model parameters.
D. Use the Amazon Bedrock Converse API and IAM roles for authentication. Pass previous messages in the request messages array to maintain conversational context. Use programming language-specific SDKs to establish consistent API interfaces.

Answer: D

Explanation: Option D is the best solution because it uses the Amazon Bedrock Converse API which is specifically designed to handle conversational context by passing previous messages in the request messages array. It uses IAM roles for consistent authentication across both environments (Lambda and EKS). It also uses the AWS SDKs (Python and JavaScript) which provide consistent API interfaces while being language-specific. This meets all requirements: same FM across environments, consistent authentication, and conversational context maintenance.

Question 12 · No. 59 · Single choice

A company upgraded its Amazon Bedrock powered foundation model (FM) that supports a multilingual customer service assistant. After the upgrade, the assistant exhibited inconsistent behavior across languages. The assistant began generating different responses in some languages when presented with identical questions. The company needs a solution to detect and address similar problems for future updates. The evaluation must be completed within 45 minutes for all supported languages. The evaluation must process at least 15,000 test conversations in parallel. The evaluation process must be fully automated and integrated into the CI/CD pipeline. The solution must block deployment if quality thresholds are not met. Which solution will meet these requirements?

A. Create a distributed traffic simulation framework that sends translation-heavy workloads to the assistant in multiple languages simultaneously. Use Amazon CloudWatch metrics to monitor latency, concurrency, and throughput. Run simulations before production releases to identify infrastructure bottlenecks.
B. Deploy the assistant in multiple AWS Regions with Amazon Route 53 latency-based routing and AWS Global Accelerator to improve global performance. Store multilingual conversation logs in Amazon S3. Perform weekly post-deployment audits to review consistency.
C. Create a pre-processing pipeline that normalizes all incoming messages into a consistent format before sending the messages to the assistant. Apply rule-based checks to flag potential hallucinations in the outputs. Focus the evaluation on the normalized text to simplify testing across languages.
D. Set up standardized multilingual test conversations with identical meaning. Run the test conversations in parallel by using Amazon Bedrock model evaluation jobs. Apply similarity and hallucination thresholds. Integrate the process into the CI/CD pipeline to block releases that fail.

Answer: D

Explanation: Option D is the correct answer because it directly addresses all the stated requirements. Amazon Bedrock model evaluation jobs are specifically designed to evaluate foundation models (FMs) at scale, support parallel processing of large test datasets (15,000+ conversations), can be automated and integrated into CI/CD pipelines, support custom quality metrics like similarity and hallucination thresholds, and can block deployments when quality thresholds are not met. Option A focuses on infrastructure performance rather than output quality. Option B is about performance optimization and weekly manual audits, not automated CI/CD integration. Option C normalizes languages away, which contradicts the need to test multilingual behavior across different languages.

Question 13 · No. 60 · Single choice

A healthcare company is developing a document management system that stores medical research papers in an Amazon S3 bucket. The company needs to build a comprehensive metadata framework that will improve search precision for a generative AI (GenAI) application that analyzes the research papers. The metadata framework must include document timestamps, author information, and research domain classifications. The solution must maintain a consistent metadata structure across all uploaded documents. The solution must give foundation models (FMs) the ability to understand document context without accessing the full content. Which solution will meet these requirements?

A. Store document timestamps in Amazon S3 system metadata. Use S3 object tags to implement domain classification. Implement custom user-defined metadata to store author information.
B. Set up S3 Object Lock with legal holds to track document timestamps. Use S3 object tags to store author information. Implement S3 access points for domain classification.
C. Use S3 Inventory reports to track document timestamps. Create S3 access points to implement domain classification. Store author information in S3 Storage Lens dashboards.
D. Use custom user-defined metadata to store author information. Use S3 Object Lock retention periods to track document timestamps. Use S3 Event Notifications to implement domain classification.

Answer: A

Explanation: Option A is the correct choice because it uses the most appropriate S3 features for each type of metadata: (1) S3 system metadata automatically tracks document timestamps (such as Last-Modified dates), providing consistent timestamp information without manual management; (2) S3 object tags are designed for categorization and classification purposes like research domain classifications, allowing easy retrieval and filtering; (3) Custom user-defined metadata is the standard AWS approach for storing custom attributes like author information, maintaining a consistent metadata structure across all documents. Together, these metadata approaches allow foundation models to understand document context by querying the metadata without needing to access the full document content. Options B, C, and D misuse AWS features: S3 Object Lock is for WORM compliance (not timestamps), S3 Inventory and Storage Lens are for storage analytics (not metadata storage), S3 access points are for access management (not classification), and S3 Event Notifications are for triggering workflows (not categorization).

Question 14 · No. 61 · Single choice

A financial services company is developing a customer service AI assistant application that uses a foundation model (FM) in Amazon Bedrock. The application must provide transparent responses by documenting reasoning and by citing sources that are used for Retrieval Augmented Generation (RAG). The application must capture comprehensive audit trails for all responses to users. The application must be able to serve up to 10,000 concurrent users and must respond to each customer inquiry within 2 seconds. Which solution will meet these requirements with the LEAST operational overhead?

A. Enable tracing for Amazon Bedrock agents. Configure structured prompts that direct the FM to provide evidence presentations. Integrate Amazon Bedrock knowledge bases with data sources to enable RAG. Configure the application to reference and cite authoritative content. Deploy the application in a Multi-AZ architecture. Use Amazon API Gateway and AWS Lambda functions to scale the application. Use Amazon CloudFront to provide low-latency delivery.
B. Enable tracing for Amazon Bedrock agents. Integrate a custom RAG pipeline with Amazon OpenSearch Service to retrieve and cite sources. Configure structured prompts to present retrieved evidence. Deploy the application behind an Amazon API Gateway REST API. Use AWS Lambda functions and Amazon CloudFront to scale the application and to provide low latency. Store logs in Amazon S3 and use AWS CloudTrail to capture audit trails.
C. Use Amazon CloudWatch to monitor latency and error rates. Embed model prompts directly in the application backend to cite sources. Store application interactions with users in Amazon RDS for audits.
D. Store generated responses and supporting evidence in an Amazon S3 bucket. Enable versioning on the bucket for audits. Use AWS Glue to catalog retrieved documents. Process the retrieved documents in Amazon Athena to generate periodic compliance reports.

Answer: A

Explanation: Option A is the best choice because it uses Amazon Bedrock's managed services (Bedrock agents with tracing and Bedrock knowledge bases) which provide the LEAST operational overhead. Here's the breakdown: (1) Tracing for Amazon Bedrock agents automatically documents the reasoning and provides audit trails; (2) Amazon Bedrock knowledge bases is a managed RAG solution that handles source citation with minimal operational burden compared to building a custom RAG pipeline (as in Option B with OpenSearch); (3) The Multi-AZ architecture with API Gateway, Lambda, and CloudFront provides automatic scaling for 10,000 concurrent users and low-latency delivery for sub-2-second response times. Option B has more operational overhead due to the custom RAG pipeline with OpenSearch. Option C lacks proper RAG integration and doesn't use Bedrock's built-in capabilities. Option D doesn't address scalability, latency, or transparent responses adequately.

Question 15 · No. 62 · Single choice

A bank is developing a generative AI (GenAI)-powered AI assistant that uses Amazon Bedrock to assist the bank's website users with account inquiries and financial guidance. The bank must ensure that the AI assistant does not reveal any personally identifiable information (PII) in customer interactions. The AI assistant must not send PII in prompts to the GenAI model. The AI assistant must not respond to customer requests to provide investment advice. The bank must collect audit logs of all customer interactions, including any images or documents that are transmitted during customer interactions. Which solution will meet these requirements with the LEAST operational effort?

A. Use Amazon Macie to detect and redact PII in user inputs and in the model responses. Apply prompt engineering techniques to force the model to avoid investment advice topics. Use AWS CloudTrail to capture conversation logs.
B. Use an AWS Lambda function and Amazon Comprehend to detect and redact PII. Use Amazon Comprehend topic modeling to prevent the AI assistant from discussing investment advice topics. Set up custom metrics in Amazon CloudWatch to capture customer conversations.
C. Configure Amazon Bedrock guardrails to apply a sensitive information policy to detect and filter PII. Set up a topic policy to ensure that the AI assistant avoids investment advice topics. Use the Converse API to log model invocations. Enable delivery and image logging to Amazon S3.
D. Use regex controls to match patterns for PII. Apply prompt engineering techniques to avoid returning PII or investment advice topics to customers. Enable model invocation logging, delivery logging, and image logging to Amazon S3.

Answer: C

Explanation: Option C is the best choice because Amazon Bedrock Guardrails is a native, managed feature specifically designed for these use cases. It provides built-in sensitive information filtering for PII detection and topic policies to block denied topics like investment advice, all without requiring custom code or external services. The Converse API natively supports invocation logging, and Bedrock's built-in image/document logging to S3 captures the required audit trails. This managed approach requires significantly less operational effort compared to Options A (using Macie, which is for S3 scanning), B (requiring Lambda functions and custom integrations), and D (using regex patterns and prompt engineering, which are less reliable and require more maintenance).

Question 16 · No. 63 · Single choice

A retail company is using Amazon Bedrock to develop a customer service AI assistant. Analysis shows that 70% of customer inquiries are simple product questions that a smaller model can effectively handle. However, 30% of inquiries are complex return policy questions that require advanced reasoning. The company wants to implement a cost-effective model selection framework to automatically route customer inquiries to appropriate models based on inquiry complexity. The framework must maintain high customer satisfaction and minimize response latency. Which solution will meet these requirements with the LEAST implementation effort?

A. Create a multi-stage architecture that uses a small foundation model (FM) to classify the complexity of each inquiry. Route simple inquiries to a smaller, more cost-effective model. Route complex inquiries to a larger, more capable model. Use AWS Lambda functions to handle the routing logic.
B. Use Amazon Bedrock intelligent prompt routing to automatically analyze inquiries. Route simple product inquiries to smaller models, and route complex return policy inquiries to more capable larger models.
C. Implement a single-model solution that uses an Amazon Bedrock mid-sized foundation model (FM) with on-demand pricing. Include special instructions in model prompts to handle both simple and complex inquiries by using the same model.
D. Create separate Amazon Bedrock endpoints for simple and complex inquiries. Implement a rule-based routing system based on keyword detection. Use on-demand pricing for the smaller model and provisioned throughput for the larger model.

Answer: B

Explanation: Amazon Bedrock Intelligent Prompt Routing is a built-in, managed feature specifically designed to automatically route prompts to different foundation models based on query complexity. It requires minimal implementation effort because the routing logic is handled natively by the Bedrock service - the company simply needs to configure the routing configuration to direct simple product inquiries to smaller, more cost-effective models and complex return policy inquiries to larger, more capable models. Options A and D require building custom routing logic with Lambda functions or rule-based keyword detection systems, which demands significantly more development and maintenance effort. Option C uses a single mid-sized model for everything, which fails to optimize costs for the 70% of simple inquiries and doesn't address the model selection framework requirement.

Question 17 · No. 64 · Single choice

A hotel company wants to enhance a legacy Java-based property management system (PMS) by adding AI capabilities. The company wants to use Amazon Bedrock Knowledge Bases to provide staff with room availability information and hotel-specific details. The solution must maintain separate access controls for each hotel that the company manages. The solution must provide room availability information in near real time and must maintain consistent performance during peak usage periods. Which solution will meet these requirements?

A. Deploy a single Amazon Bedrock knowledge base that contains combined data for all hotels. Configure AWS Lambda functions to synchronize data from each hotel's PMS database through direct API connections. Implement AWS CloudTrail logging with hotel-specific filters to audit access logs for each hotel's data.
B. Create an Amazon EventBridge rule for each hotel that is invoked by changes to the PMS database for each hotel. Configure the rule to send updates to a centralized Amazon Bedrock knowledge base in a management AWS account. Configure resource-based policies to enforce hotel-specific access controls for hotel staff.
C. Implement one Amazon Bedrock knowledge base for each hotel in a multi-account structure. Use direct data ingestion to provide real-time room availability information. Schedule regular synchronization for less critical information.
D. Build a centralized Amazon Bedrock agent that uses multiple knowledge bases. Implement AWS IAM Identity Center with hotel-specific permission sets to control hotel staff data access.

Answer: C

Explanation: Option C best meets all the requirements. Implementing one Amazon Bedrock knowledge base per hotel in a multi-account structure provides strong isolation for separate access controls, ensuring each hotel's data and permissions are kept independent. Direct data ingestion supports near real-time room availability information, which is critical for staff operations. The distributed architecture (one knowledge base per hotel) ensures consistent performance during peak usage periods, as the load is spread across multiple resources rather than concentrated on a single centralized system. Scheduled synchronization for less critical information optimizes costs while maintaining data freshness where it matters most.

Question 18 · No. 65 · Single choice

A company wants to create an annual rewards program for its customers. The rewards that customers earn vary based on different parameters such as the categories of the items ordered and the customers' purchase history. The company needs a generative AI (GenAI) solution that uses three Amazon Bedrock agents to help customers during online catalog browsing. The agents must use knowledge bases and action groups to handle the search, recommendation, and order modules. The modules must operate sequentially. An AWS Lambda function must calculate estimated rewards for each recommended item. The solution must provide graceful degradation during service disruptions. Which solution will meet these requirements with the MOST operational efficiency?

A. Define an Amazon API Gateway REST API behind each agent. Create a second Lambda function to orchestrate the calls to the agents and the rewards Lambda function. Configure the second Lambda function with a retry/fallback mechanism.
B. Create an AWS Step Functions state machine with four tasks that run the agents and the rewards Lambda function. Set up retry and catch branches for each of the task steps.
C. Configure each agent with a separate retry/fallback mechanism. Create a second Lambda function to orchestrate the calls to the agents and the rewards Lambda function. Define an Amazon API Gateway REST API behind the second Lambda function.
D. Create a second Lambda function to orchestrate the calls to the agents and the rewards Lambda function. Create an AWS Step Functions state machine with one task that runs the second Lambda function. Set up retry and catch branches for the task step.

Answer: B

Explanation: Option B is the most operationally efficient solution because AWS Step Functions is purpose-built for orchestrating multiple sequential tasks. The state machine can natively handle the three Bedrock agents and the rewards Lambda function as four separate tasks, executing them sequentially as required. Step Functions provides built-in retry and catch mechanisms that directly address the graceful degradation requirement without needing additional custom code. This eliminates the need for extra orchestration Lambda functions (Options A, C, D) or multiple API Gateways (Option A), and avoids the complexity of wrapping a Lambda orchestrator inside Step Functions (Option D). As a managed service, Step Functions reduces operational overhead by handling state management, error handling, and retry logic automatically, making it the most operationally efficient choice.

Question 19 · No. 66 · Single choice

A company uses Amazon Bedrock to develop an AI assistant to provide customer support. Analysis shows that 40% of customer queries use varied phrasing or wording to ask the same questions. The company wants a solution to reduce redundant model calls. The solution must ensure that semantically equivalent questions receive consistent answers. The solution must ensure low latency. Which solution will meet these requirements?

A. Deploy an Amazon DynamoDB Accelerator (DAX) cluster as an in-memory cache. Specify the query text as the partition key and the model response text as the sort key. Query the cache by using a filter expression with the LIKE operator.
B. Use Amazon Bedrock to generate embeddings from customer queries. Use Amazon MemoryDB for Valkey to store hash sets of vector embeddings and model responses. Use a RANGE query to find similar queries and their responses.
C. Deploy Amazon OpenSearch Service that has k-nearest neighbor (k-NN) capabilities to store query-response text pairs. Use an approximate k-NN technique to find similar queries and their responses.
D. Create a caching solution by using Amazon DynamoDB to create a global secondary index on the normalized query text. Apply stemming to incoming queries. Query the index of cached customer queries.

Answer: C

Explanation: C is correct because Amazon OpenSearch Service with k-nearest neighbor (k-NN) can store query-response pairs and use approximate k-NN search to identify semantically similar queries. This supports varied phrasing, consistent cached responses, reduced model calls, and low-latency retrieval. DAX with LIKE and DynamoDB normalization do not provide semantic matching, while the MemoryDB RANGE query does not perform effective vector similarity search.

Question 20 · No. 67 · Single choice

A company is using Amazon Bedrock and Anthropic Claude 3 Haiku to develop an AI assistant. The AI assistant normally processes 10,000 requests each hour but experiences surges of up 30,000 requests each hour during peak usage periods. The AI assistant must respond within 2 seconds while operating across multiple AWS Regions. The company observes that during peak usage periods, the AI assistant experiences throughput bottlenecks that cause increased latency and occasional request timeouts. The company must resolve the performance issues. Which solution will meet this requirement?

A. Purchase provisioned throughput and sufficient model units (MUs) in a single Region. Configure the application to retry failed requests with exponential backoff.
B. Implement token batching to reduce API overhead. Use cross-Region inference profiles to automatically distribute traffic across available Regions.
C. Set up auto scaling AWS Lambda functions in each Region. Implement client-side round-robin request distribution. Purchase one model unit (MU) of provisioned throughput as a backup.
D. Implement batch inference for all requests by using Amazon S3 buckets across multiple Regions. Use Amazon SQS to set up an asynchronous retrieval process.

Answer: B

Explanation: Option B is the best solution because it addresses both the throughput bottleneck and the multi-Region requirement. Token batching reduces API overhead by combining multiple tokens/requests into fewer API calls, which directly helps with throughput. Cross-Region inference profiles are a native Amazon Bedrock feature that automatically distributes traffic across multiple Regions, which is essential when the company needs to handle surges from 10,000 to 30,000 requests per hour. This solution maintains the 2-second response time requirement because it handles requests synchronously while distributing load. Option A fails because a single Region cannot effectively handle 30,000 peak requests, and exponential backoff would add latency that violates the 2-second SLA. Option C is insufficient because one model unit as backup cannot handle 30,000 peak requests, and client-side round-robin lacks intelligent failover. Option D is incorrect because batch inference with SQS is asynchronous and would not meet the 2-second real-time response requirement.

Prev 123 4 5 Next