Amazon AWS AIP-C01 RAG and Retrieval Practice Test

 

Topic 05 focuses on Retrieval-Augmented Generation and Retrieval Mechanisms for the AWS Certified Generative AI Developer – Professional certification and the AIP-C01 exam, using Amazon Bedrock and AWS generative AI services and architecture scenarios where relevant. For broader exam preparation, review the AWS Certified Generative AI Developer – Professional AIP-C01 Exam Dumps page. Each question includes a concise explanation of the correct answer and the technical reason the other choices are incorrect.

Question 1

The contact-center transformation team is prioritizing a requirement to split uniform documents into predictable token or character windows for retrieval. Which implementation is most appropriate?

1. hierarchical chunking
2. Bedrock reranker model
3. OpenSearch semantic vector retrieval
4. fixed-size chunking

Correct Answer: D

 

Correct Answer

Answer D is correct because fixed-size chunking is designed to split uniform documents into predictable token or character windows for retrieval. It creates chunks of approximately consistent size and overlap, which is simple but may ignore semantic document boundaries.

Incorrect Answers

Answer A is incorrect because hierarchical chunking is primarily used to preserve sections and parent-child context when documents have meaningful structure, which is a different requirement from the one being tested.

Answer B is incorrect because Bedrock reranker model is primarily used to reorder initially retrieved passages so the most contextually relevant items appear first, which is a different requirement from the one being tested.

Answer C is incorrect because OpenSearch semantic vector retrieval is primarily used to perform nearest-neighbor search over embeddings with scalable search infrastructure, which is a different requirement from the one being tested.

 

Question 2

The customer analytics platform has a design goal to preserve sections and parent-child context when documents have meaningful structure. What should the team choose?

1. query expansion
2. Amazon Titan embedding model
3. hierarchical chunking
4. Aurora pgvector retrieval

Correct Answer: C

 

Correct Answer

Answer C is correct because hierarchical chunking is designed to preserve sections and parent-child context when documents have meaningful structure. It segments content according to headings or hierarchy so retrieved passages can retain their broader parent context.

Incorrect Answers

Answer A is incorrect because query expansion is primarily used to improve recall when a user query is too narrow or uses alternate terminology, which is a different requirement from the one being tested.

Answer B is incorrect because Amazon Titan embedding model is primarily used to convert text into vector representations for semantic retrieval, which is a different requirement from the one being tested.

Answer D is incorrect because Aurora pgvector retrieval is primarily used to perform vector similarity search alongside relational filtering and transactional data, which is a different requirement from the one being tested.

 

Question 3

The enterprise finance automation team is prioritizing a requirement to convert text into vector representations for semantic retrieval. Which implementation is most appropriate?

1. query decomposition and transformation
2. batched embedding generation with Lambda
3. Bedrock Knowledge Bases managed retrieval
4. Amazon Titan embedding model

Correct Answer: D

 

Correct Answer

Answer D is correct because Amazon Titan embedding model is designed to convert text into vector representations for semantic retrieval. It generates embeddings whose dimensionality and domain performance can be evaluated against the retrieval use case.

Incorrect Answers

Answer A is incorrect because query decomposition and transformation is primarily used to break a complex request into retrievable subquestions or normalize it for better search, which is a different requirement from the one being tested.

Answer B is incorrect because batched embedding generation with Lambda is primarily used to reduce per-record overhead when many small documents need embeddings, which is a different requirement from the one being tested.

Answer C is incorrect because Bedrock Knowledge Bases managed retrieval is primarily used to reduce custom retrieval plumbing for a standard RAG application, which is a different requirement from the one being tested.

 

Question 4

The risk and controls team needs an architecture that can reduce per-record overhead when many small documents need embeddings. Which choice best meets that need?

1. OpenSearch semantic vector retrieval
2. hybrid search
3. batched embedding generation with Lambda
4. standardized retrieval interface with MCP or function calling

Correct Answer: C

 

Correct Answer

Answer C is correct because batched embedding generation with Lambda is designed to reduce per-record overhead when many small documents need embeddings. It groups eligible items into controlled batches before invoking embedding generation rather than making one invocation per tiny record.

Incorrect Answers

Answer A is incorrect because OpenSearch semantic vector retrieval is primarily used to perform nearest-neighbor search over embeddings with scalable search infrastructure, which is a different requirement from the one being tested.

Answer B is incorrect because hybrid search is primarily used to combine semantic similarity with lexical keyword signals, which is a different requirement from the one being tested.

Answer D is incorrect because standardized retrieval interface with MCP or function calling is primarily used to let agents and foundation models access retrieval tools through a consistent contract, which is a different requirement from the one being tested.

 

Question 5

The enterprise search architecture group must perform nearest-neighbor search over embeddings with scalable search infrastructure. Which approach is the strongest fit?

1. OpenSearch semantic vector retrieval
2. Aurora pgvector retrieval
3. fixed-size chunking
4. Bedrock reranker model

Correct Answer: A

 

Correct Answer

Answer A is correct because OpenSearch semantic vector retrieval is designed to perform nearest-neighbor search over embeddings with scalable search infrastructure. It queries vector indexes to return semantically similar passages even when exact keywords differ.

Incorrect Answers

Answer B is incorrect because Aurora pgvector retrieval is primarily used to perform vector similarity search alongside relational filtering and transactional data, which is a different requirement from the one being tested.

Answer C is incorrect because fixed-size chunking is primarily used to split uniform documents into predictable token or character windows for retrieval, which is a different requirement from the one being tested.

Answer D is incorrect because Bedrock reranker model is primarily used to reorder initially retrieved passages so the most contextually relevant items appear first, which is a different requirement from the one being tested.

 

Question 6

The insurance automation team needs an architecture that can perform vector similarity search alongside relational filtering and transactional data. Which choice best meets that need?

1. hierarchical chunking
2. Aurora pgvector retrieval
3. Bedrock Knowledge Bases managed retrieval
4. query expansion

Correct Answer: B

 

Correct Answer

Answer B is correct because Aurora pgvector retrieval is designed to perform vector similarity search alongside relational filtering and transactional data. It uses postgresql vector operators and indexes within aurora to combine semantic and relational query criteria.

Incorrect Answers

Answer A is incorrect because hierarchical chunking is primarily used to preserve sections and parent-child context when documents have meaningful structure, which is a different requirement from the one being tested.

Answer C is incorrect because Bedrock Knowledge Bases managed retrieval is primarily used to reduce custom retrieval plumbing for a standard RAG application, which is a different requirement from the one being tested.

Answer D is incorrect because query expansion is primarily used to improve recall when a user query is too narrow or uses alternate terminology, which is a different requirement from the one being tested.

 

Question 7

The procurement automation group has a design goal to reduce custom retrieval plumbing for a standard RAG application. What should the team choose?

1. hybrid search
2. Amazon Titan embedding model
3. Bedrock Knowledge Bases managed retrieval
4. query decomposition and transformation

Correct Answer: C

 

Correct Answer

Answer C is correct because Bedrock Knowledge Bases managed retrieval is designed to reduce custom retrieval plumbing for a standard RAG application. It provides managed retrieval orchestration between data sources, embeddings, vector stores, and Bedrock foundation models.

Incorrect Answers

Answer A is incorrect because hybrid search is primarily used to combine semantic similarity with lexical keyword signals, which is a different requirement from the one being tested.

Answer B is incorrect because Amazon Titan embedding model is primarily used to convert text into vector representations for semantic retrieval, which is a different requirement from the one being tested.

Answer D is incorrect because query decomposition and transformation is primarily used to break a complex request into retrievable subquestions or normalize it for better search, which is a different requirement from the one being tested.

 

Question 8

The observability engineering group is prioritizing a requirement to combine semantic similarity with lexical keyword signals. Which implementation is most appropriate?

1. hybrid search
2. batched embedding generation with Lambda
3. standardized retrieval interface with MCP or function calling
4. Bedrock reranker model

Correct Answer: A

 

Correct Answer

Answer A is correct because hybrid search is designed to combine semantic similarity with lexical keyword signals. It merges or scores vector and keyword results so exact terms and semantic meaning both contribute to retrieval relevance.

Incorrect Answers

Answer B is incorrect because batched embedding generation with Lambda is primarily used to reduce per-record overhead when many small documents need embeddings, which is a different requirement from the one being tested.

Answer C is incorrect because standardized retrieval interface with MCP or function calling is primarily used to let agents and foundation models access retrieval tools through a consistent contract, which is a different requirement from the one being tested.

Answer D is incorrect because Bedrock reranker model is primarily used to reorder initially retrieved passages so the most contextually relevant items appear first, which is a different requirement from the one being tested.

 

Question 9

The managed services operations team has a design goal to reorder initially retrieved passages so the most contextually relevant items appear first. What should the team choose?

1. query expansion
2. fixed-size chunking
3. OpenSearch semantic vector retrieval
4. Bedrock reranker model

Correct Answer: D

 

Correct Answer

Answer D is correct because Bedrock reranker model is designed to reorder initially retrieved passages so the most contextually relevant items appear first. It applies a second-stage relevance model to candidate results before context is sent to the foundation model.

Incorrect Answers

Answer A is incorrect because query expansion is primarily used to improve recall when a user query is too narrow or uses alternate terminology, which is a different requirement from the one being tested.

Answer B is incorrect because fixed-size chunking is primarily used to split uniform documents into predictable token or character windows for retrieval, which is a different requirement from the one being tested.

Answer C is incorrect because OpenSearch semantic vector retrieval is primarily used to perform nearest-neighbor search over embeddings with scalable search infrastructure, which is a different requirement from the one being tested.

 

Question 10

The service management platform needs an architecture that can improve recall when a user query is too narrow or uses alternate terminology. Which choice best meets that need?

1. Aurora pgvector retrieval
2. hierarchical chunking
3. query decomposition and transformation
4. query expansion

Correct Answer: D

 

Correct Answer

Answer D is correct because query expansion is designed to improve recall when a user query is too narrow or uses alternate terminology. It adds related terms or reformulations so retrieval can find relevant content that the original wording might miss.

Incorrect Answers

Answer A is incorrect because Aurora pgvector retrieval is primarily used to perform vector similarity search alongside relational filtering and transactional data, which is a different requirement from the one being tested.

Answer B is incorrect because hierarchical chunking is primarily used to preserve sections and parent-child context when documents have meaningful structure, which is a different requirement from the one being tested.

Answer C is incorrect because query decomposition and transformation is primarily used to break a complex request into retrievable subquestions or normalize it for better search, which is a different requirement from the one being tested.

 

Question 11

The digital experience platform must break a complex request into retrievable subquestions or normalize it for better search. Which approach is the strongest fit?

1. Bedrock Knowledge Bases managed retrieval
2. query decomposition and transformation
3. Amazon Titan embedding model
4. standardized retrieval interface with MCP or function calling

Correct Answer: B

 

Correct Answer

Answer B is correct because query decomposition and transformation is designed to break a complex request into retrievable subquestions or normalize it for better search. It uses logic or models to rewrite, split, or sequence retrieval queries before collecting context.

Incorrect Answers

Answer A is incorrect because Bedrock Knowledge Bases managed retrieval is primarily used to reduce custom retrieval plumbing for a standard RAG application, which is a different requirement from the one being tested.

Answer C is incorrect because Amazon Titan embedding model is primarily used to convert text into vector representations for semantic retrieval, which is a different requirement from the one being tested.

Answer D is incorrect because standardized retrieval interface with MCP or function calling is primarily used to let agents and foundation models access retrieval tools through a consistent contract, which is a different requirement from the one being tested.

 

Question 12

The cloud governance office must let agents and foundation models access retrieval tools through a consistent contract. Which approach is the strongest fit?

1. standardized retrieval interface with MCP or function calling
2. fixed-size chunking
3. batched embedding generation with Lambda
4. hybrid search

Correct Answer: A

 

Correct Answer

Answer A is correct because standardized retrieval interface with MCP or function calling is designed to let agents and foundation models access retrieval tools through a consistent contract. It exposes vector search as a well-defined callable tool instead of embedding store-specific logic throughout the application.

Incorrect Answers

Answer B is incorrect because fixed-size chunking is primarily used to split uniform documents into predictable token or character windows for retrieval, which is a different requirement from the one being tested.

Answer C is incorrect because batched embedding generation with Lambda is primarily used to reduce per-record overhead when many small documents need embeddings, which is a different requirement from the one being tested.

Answer D is incorrect because hybrid search is primarily used to combine semantic similarity with lexical keyword signals, which is a different requirement from the one being tested.

 

Question 13

Within the application security office’s architecture, which capability matches this technical description: creates chunks of approximately consistent size and overlap, which is simple but may ignore semantic document boundaries?

1. hierarchical chunking
2. OpenSearch semantic vector retrieval
3. fixed-size chunking
4. Bedrock reranker model

Correct Answer: C

 

Correct Answer

Answer C is correct because the description directly matches fixed-size chunking. fixed-size chunking creates chunks of approximately consistent size and overlap, which is simple but may ignore semantic document boundaries.

Incorrect Answers

Answer A is incorrect because hierarchical chunking is primarily used to preserve sections and parent-child context when documents have meaningful structure, which is a different requirement from the one being tested.

Answer B is incorrect because OpenSearch semantic vector retrieval is primarily used to perform nearest-neighbor search over embeddings with scalable search infrastructure, which is a different requirement from the one being tested.

Answer D is incorrect because Bedrock reranker model is primarily used to reorder initially retrieved passages so the most contextually relevant items appear first, which is a different requirement from the one being tested.

 

Question 14

Within the quality engineering group’s architecture, which capability matches this technical description: segments content according to headings or hierarchy so retrieved passages can retain their broader parent context?

1. Aurora pgvector retrieval
2. query expansion
3. Amazon Titan embedding model
4. hierarchical chunking

Correct Answer: D

 

Correct Answer

Answer D is correct because the description directly matches hierarchical chunking. hierarchical chunking segments content according to headings or hierarchy so retrieved passages can retain their broader parent context.

Incorrect Answers

Answer A is incorrect because Aurora pgvector retrieval is primarily used to perform vector similarity search alongside relational filtering and transactional data, which is a different requirement from the one being tested.

Answer B is incorrect because query expansion is primarily used to improve recall when a user query is too narrow or uses alternate terminology, which is a different requirement from the one being tested.

Answer C is incorrect because Amazon Titan embedding model is primarily used to convert text into vector representations for semantic retrieval, which is a different requirement from the one being tested.

 

Question 15

The media processing service documents this GenAI behavior: generates embeddings whose dimensionality and domain performance can be evaluated against the retrieval use case. Which capability matches it?

1. Bedrock Knowledge Bases managed retrieval
2. query decomposition and transformation
3. batched embedding generation with Lambda
4. Amazon Titan embedding model

Correct Answer: D

 

Correct Answer

Answer D is correct because the description directly matches Amazon Titan embedding model. Amazon Titan embedding model generates embeddings whose dimensionality and domain performance can be evaluated against the retrieval use case.

Incorrect Answers

Answer A is incorrect because Bedrock Knowledge Bases managed retrieval is primarily used to reduce custom retrieval plumbing for a standard RAG application, which is a different requirement from the one being tested.

Answer B is incorrect because query decomposition and transformation is primarily used to break a complex request into retrievable subquestions or normalize it for better search, which is a different requirement from the one being tested.

Answer C is incorrect because batched embedding generation with Lambda is primarily used to reduce per-record overhead when many small documents need embeddings, which is a different requirement from the one being tested.

 

Question 16

The multinational architecture team needs to identify a capability with this behavior: groups eligible items into controlled batches before invoking embedding generation rather than making one invocation per tiny record. What is the best match?

1. batched embedding generation with Lambda
2. hybrid search
3. standardized retrieval interface with MCP or function calling
4. OpenSearch semantic vector retrieval

Correct Answer: A

 

Correct Answer

Answer A is correct because the description directly matches batched embedding generation with Lambda. batched embedding generation with Lambda groups eligible items into controlled batches before invoking embedding generation rather than making one invocation per tiny record.

Incorrect Answers

Answer B is incorrect because hybrid search is primarily used to combine semantic similarity with lexical keyword signals, which is a different requirement from the one being tested.

Answer C is incorrect because standardized retrieval interface with MCP or function calling is primarily used to let agents and foundation models access retrieval tools through a consistent contract, which is a different requirement from the one being tested.

Answer D is incorrect because OpenSearch semantic vector retrieval is primarily used to perform nearest-neighbor search over embeddings with scalable search infrastructure, which is a different requirement from the one being tested.

 

Question 17

Within the platform reliability team’s architecture, which capability matches this technical description: queries vector indexes to return semantically similar passages even when exact keywords differ?

1. Bedrock reranker model
2. OpenSearch semantic vector retrieval
3. fixed-size chunking
4. Aurora pgvector retrieval

Correct Answer: B

 

Correct Answer

Answer B is correct because the description directly matches OpenSearch semantic vector retrieval. OpenSearch semantic vector retrieval queries vector indexes to return semantically similar passages even when exact keywords differ.

Incorrect Answers

Answer A is incorrect because Bedrock reranker model is primarily used to reorder initially retrieved passages so the most contextually relevant items appear first, which is a different requirement from the one being tested.

Answer C is incorrect because fixed-size chunking is primarily used to split uniform documents into predictable token or character windows for retrieval, which is a different requirement from the one being tested.

Answer D is incorrect because Aurora pgvector retrieval is primarily used to perform vector similarity search alongside relational filtering and transactional data, which is a different requirement from the one being tested.

 

Question 18

The enterprise search program describes a component that uses PostgreSQL vector operators and indexes within Aurora to combine semantic and relational query criteria. Which capability is being described?

1. hierarchical chunking
2. Bedrock Knowledge Bases managed retrieval
3. Aurora pgvector retrieval
4. query expansion

Correct Answer: C

 

Correct Answer

Answer C is correct because the description directly matches Aurora pgvector retrieval. Aurora pgvector retrieval uses PostgreSQL vector operators and indexes within Aurora to combine semantic and relational query criteria.

Incorrect Answers

Answer A is incorrect because hierarchical chunking is primarily used to preserve sections and parent-child context when documents have meaningful structure, which is a different requirement from the one being tested.

Answer B is incorrect because Bedrock Knowledge Bases managed retrieval is primarily used to reduce custom retrieval plumbing for a standard RAG application, which is a different requirement from the one being tested.

Answer D is incorrect because query expansion is primarily used to improve recall when a user query is too narrow or uses alternate terminology, which is a different requirement from the one being tested.

 

Question 19

The workflow modernization program describes a component that provides managed retrieval orchestration between data sources, embeddings, vector stores, and Bedrock foundation models. Which capability is being described?

1. Amazon Titan embedding model
2. Bedrock Knowledge Bases managed retrieval
3. hybrid search
4. query decomposition and transformation

Correct Answer: B

 

Correct Answer

Answer B is correct because the description directly matches Bedrock Knowledge Bases managed retrieval. Bedrock Knowledge Bases managed retrieval provides managed retrieval orchestration between data sources, embeddings, vector stores, and Bedrock foundation models.

Incorrect Answers

Answer A is incorrect because Amazon Titan embedding model is primarily used to convert text into vector representations for semantic retrieval, which is a different requirement from the one being tested.

Answer C is incorrect because hybrid search is primarily used to combine semantic similarity with lexical keyword signals, which is a different requirement from the one being tested.

Answer D is incorrect because query decomposition and transformation is primarily used to break a complex request into retrievable subquestions or normalize it for better search, which is a different requirement from the one being tested.

 

Question 20

The medical research assistant needs to identify a capability with this behavior: merges or scores vector and keyword results so exact terms and semantic meaning both contribute to retrieval relevance. What is the best match?

1. standardized retrieval interface with MCP or function calling
2. Bedrock reranker model
3. hybrid search
4. batched embedding generation with Lambda

Correct Answer: C

 

Correct Answer

Answer C is correct because the description directly matches hybrid search. hybrid search merges or scores vector and keyword results so exact terms and semantic meaning both contribute to retrieval relevance.

Incorrect Answers

Answer A is incorrect because standardized retrieval interface with MCP or function calling is primarily used to let agents and foundation models access retrieval tools through a consistent contract, which is a different requirement from the one being tested.

Answer B is incorrect because Bedrock reranker model is primarily used to reorder initially retrieved passages so the most contextually relevant items appear first, which is a different requirement from the one being tested.

Answer D is incorrect because batched embedding generation with Lambda is primarily used to reduce per-record overhead when many small documents need embeddings, which is a different requirement from the one being tested.

 

Question 21

The regional cloud architecture group describes a component that applies a second-stage relevance model to candidate results before context is sent to the foundation model. Which capability is being described?

1. OpenSearch semantic vector retrieval
2. query expansion
3. Bedrock reranker model
4. fixed-size chunking

Correct Answer: C

 

Correct Answer

Answer C is correct because the description directly matches Bedrock reranker model. Bedrock reranker model applies a second-stage relevance model to candidate results before context is sent to the foundation model.

Incorrect Answers

Answer A is incorrect because OpenSearch semantic vector retrieval is primarily used to perform nearest-neighbor search over embeddings with scalable search infrastructure, which is a different requirement from the one being tested.

Answer B is incorrect because query expansion is primarily used to improve recall when a user query is too narrow or uses alternate terminology, which is a different requirement from the one being tested.

Answer D is incorrect because fixed-size chunking is primarily used to split uniform documents into predictable token or character windows for retrieval, which is a different requirement from the one being tested.

 

Question 22

The retail recommendation team documents this GenAI behavior: adds related terms or reformulations so retrieval can find relevant content that the original wording might miss. Which capability matches it?

1. Aurora pgvector retrieval
2. query expansion
3. hierarchical chunking
4. query decomposition and transformation

Correct Answer: B

 

Correct Answer

Answer B is correct because the description directly matches query expansion. query expansion adds related terms or reformulations so retrieval can find relevant content that the original wording might miss.

Incorrect Answers

Answer A is incorrect because Aurora pgvector retrieval is primarily used to perform vector similarity search alongside relational filtering and transactional data, which is a different requirement from the one being tested.

Answer C is incorrect because hierarchical chunking is primarily used to preserve sections and parent-child context when documents have meaningful structure, which is a different requirement from the one being tested.

Answer D is incorrect because query decomposition and transformation is primarily used to break a complex request into retrievable subquestions or normalize it for better search, which is a different requirement from the one being tested.

 

Question 23

The telecommunications service describes a component that uses logic or models to rewrite, split, or sequence retrieval queries before collecting context. Which capability is being described?

1. Amazon Titan embedding model
2. query decomposition and transformation
3. Bedrock Knowledge Bases managed retrieval
4. standardized retrieval interface with MCP or function calling

Correct Answer: B

 

Correct Answer

Answer B is correct because the description directly matches query decomposition and transformation. query decomposition and transformation uses logic or models to rewrite, split, or sequence retrieval queries before collecting context.

Incorrect Answers

Answer A is incorrect because Amazon Titan embedding model is primarily used to convert text into vector representations for semantic retrieval, which is a different requirement from the one being tested.

Answer C is incorrect because Bedrock Knowledge Bases managed retrieval is primarily used to reduce custom retrieval plumbing for a standard RAG application, which is a different requirement from the one being tested.

Answer D is incorrect because standardized retrieval interface with MCP or function calling is primarily used to let agents and foundation models access retrieval tools through a consistent contract, which is a different requirement from the one being tested.

 

Question 24

Within the marketing content platform’s architecture, which capability matches this technical description: exposes vector search as a well-defined callable tool instead of embedding store-specific logic throughout the application?

1. hybrid search
2. standardized retrieval interface with MCP or function calling
3. batched embedding generation with Lambda
4. fixed-size chunking

Correct Answer: B

 

Correct Answer

Answer B is correct because the description directly matches standardized retrieval interface with MCP or function calling. standardized retrieval interface with MCP or function calling exposes vector search as a well-defined callable tool instead of embedding store-specific logic throughout the application.

Incorrect Answers

Answer A is incorrect because hybrid search is primarily used to combine semantic similarity with lexical keyword signals, which is a different requirement from the one being tested.

Answer C is incorrect because batched embedding generation with Lambda is primarily used to reduce per-record overhead when many small documents need embeddings, which is a different requirement from the one being tested.

Answer D is incorrect because fixed-size chunking is primarily used to split uniform documents into predictable token or character windows for retrieval, which is a different requirement from the one being tested.

 

Question 25

The public-sector application team is considering fixed-size chunking. What problem is this choice primarily meant to solve?

1. Split uniform documents into predictable token or character windows for retrieval
2. Reorder initially retrieved passages so the most contextually relevant items appear first
3. Preserve sections and parent-child context when documents have meaningful structure
4. Perform nearest-neighbor search over embeddings with scalable search infrastructure

Correct Answer: A

 

Correct Answer

Answer A is correct because fixed-size chunking is specifically used to split uniform documents into predictable token or character windows for retrieval. It creates chunks of approximately consistent size and overlap, which is simple but may ignore semantic document boundaries.

Incorrect Answers

Answer B is incorrect because that requirement aligns with Bedrock reranker model, not fixed-size chunking.

Answer C is incorrect because that requirement aligns with hierarchical chunking, not fixed-size chunking.

Answer D is incorrect because that requirement aligns with OpenSearch semantic vector retrieval, not fixed-size chunking.

 

Question 26

Within the internal AI platform team’s design, the team highlights hierarchical chunking. Which need does that component address?

1. Improve recall when a user query is too narrow or uses alternate terminology
2. Preserve sections and parent-child context when documents have meaningful structure
3. Convert text into vector representations for semantic retrieval
4. Perform vector similarity search alongside relational filtering and transactional data

Correct Answer: B

 

Correct Answer

Answer B is correct because hierarchical chunking is specifically used to preserve sections and parent-child context when documents have meaningful structure. It segments content according to headings or hierarchy so retrieved passages can retain their broader parent context.

Incorrect Answers

Answer A is incorrect because that requirement aligns with query expansion, not hierarchical chunking.

Answer C is incorrect because that requirement aligns with Amazon Titan embedding model, not hierarchical chunking.

Answer D is incorrect because that requirement aligns with Aurora pgvector retrieval, not hierarchical chunking.

 

Question 27

The data privacy office includes Amazon Titan embedding model in its architecture. What is its primary role?

1. Convert text into vector representations for semantic retrieval
2. Reduce custom retrieval plumbing for a standard RAG application
3. Break a complex request into retrievable subquestions or normalize it for better search
4. Reduce per-record overhead when many small documents need embeddings

Correct Answer: A

 

Correct Answer

Answer A is correct because Amazon Titan embedding model is specifically used to convert text into vector representations for semantic retrieval. It generates embeddings whose dimensionality and domain performance can be evaluated against the retrieval use case.

Incorrect Answers

Answer B is incorrect because that requirement aligns with Bedrock Knowledge Bases managed retrieval, not Amazon Titan embedding model.

Answer C is incorrect because that requirement aligns with query decomposition and transformation, not Amazon Titan embedding model.

Answer D is incorrect because that requirement aligns with batched embedding generation with Lambda, not Amazon Titan embedding model.

 

Question 28

The enterprise integration office plans to adopt batched embedding generation with Lambda. Which outcome should drive that decision?

1. Let agents and foundation models access retrieval tools through a consistent contract
2. Combine semantic similarity with lexical keyword signals
3. Perform nearest-neighbor search over embeddings with scalable search infrastructure
4. Reduce per-record overhead when many small documents need embeddings

Correct Answer: D

 

Correct Answer

Answer D is correct because batched embedding generation with Lambda is specifically used to reduce per-record overhead when many small documents need embeddings. It groups eligible items into controlled batches before invoking embedding generation rather than making one invocation per tiny record.

Incorrect Answers

Answer A is incorrect because that requirement aligns with standardized retrieval interface with MCP or function calling, not batched embedding generation with Lambda.

Answer B is incorrect because that requirement aligns with hybrid search, not batched embedding generation with Lambda.

Answer C is incorrect because that requirement aligns with OpenSearch semantic vector retrieval, not batched embedding generation with Lambda.

 

Question 29

The corporate knowledge assistant includes OpenSearch semantic vector retrieval in its architecture. What is its primary role?

1. Perform nearest-neighbor search over embeddings with scalable search infrastructure
2. Perform vector similarity search alongside relational filtering and transactional data
3. Split uniform documents into predictable token or character windows for retrieval
4. Reorder initially retrieved passages so the most contextually relevant items appear first

Correct Answer: A

 

Correct Answer

Answer A is correct because OpenSearch semantic vector retrieval is specifically used to perform nearest-neighbor search over embeddings with scalable search infrastructure. It queries vector indexes to return semantically similar passages even when exact keywords differ.

Incorrect Answers

Answer B is incorrect because that requirement aligns with Aurora pgvector retrieval, not OpenSearch semantic vector retrieval.

Answer C is incorrect because that requirement aligns with fixed-size chunking, not OpenSearch semantic vector retrieval.

Answer D is incorrect because that requirement aligns with Bedrock reranker model, not OpenSearch semantic vector retrieval.

 

Question 30

The enterprise data marketplace plans to adopt Aurora pgvector retrieval. Which outcome should drive that decision?

1. Perform vector similarity search alongside relational filtering and transactional data
2. Improve recall when a user query is too narrow or uses alternate terminology
3. Preserve sections and parent-child context when documents have meaningful structure
4. Reduce custom retrieval plumbing for a standard RAG application

Correct Answer: A

 

Correct Answer

Answer A is correct because Aurora pgvector retrieval is specifically used to perform vector similarity search alongside relational filtering and transactional data. It uses postgresql vector operators and indexes within aurora to combine semantic and relational query criteria.

Incorrect Answers

Answer B is incorrect because that requirement aligns with query expansion, not Aurora pgvector retrieval.

Answer C is incorrect because that requirement aligns with hierarchical chunking, not Aurora pgvector retrieval.

Answer D is incorrect because that requirement aligns with Bedrock Knowledge Bases managed retrieval, not Aurora pgvector retrieval.

 

Question 31

The case management platform has proposed Bedrock Knowledge Bases managed retrieval for its design. Which requirement best justifies it?

1. Convert text into vector representations for semantic retrieval
2. Combine semantic similarity with lexical keyword signals
3. Reduce custom retrieval plumbing for a standard RAG application
4. Break a complex request into retrievable subquestions or normalize it for better search

Correct Answer: C

 

Correct Answer

Answer C is correct because Bedrock Knowledge Bases managed retrieval is specifically used to reduce custom retrieval plumbing for a standard RAG application. It provides managed retrieval orchestration between data sources, embeddings, vector stores, and Bedrock foundation models.

Incorrect Answers

Answer A is incorrect because that requirement aligns with Amazon Titan embedding model, not Bedrock Knowledge Bases managed retrieval.

Answer B is incorrect because that requirement aligns with hybrid search, not Bedrock Knowledge Bases managed retrieval.

Answer D is incorrect because that requirement aligns with query decomposition and transformation, not Bedrock Knowledge Bases managed retrieval.

 

Question 32

Why would the legal research platform introduce hybrid search into the GenAI architecture?

1. Combine semantic similarity with lexical keyword signals
2. Reorder initially retrieved passages so the most contextually relevant items appear first
3. Let agents and foundation models access retrieval tools through a consistent contract
4. Reduce per-record overhead when many small documents need embeddings

Correct Answer: A

 

Correct Answer

Answer A is correct because hybrid search is specifically used to combine semantic similarity with lexical keyword signals. It merges or scores vector and keyword results so exact terms and semantic meaning both contribute to retrieval relevance.

Incorrect Answers

Answer B is incorrect because that requirement aligns with Bedrock reranker model, not hybrid search.

Answer C is incorrect because that requirement aligns with standardized retrieval interface with MCP or function calling, not hybrid search.

Answer D is incorrect because that requirement aligns with batched embedding generation with Lambda, not hybrid search.

 

Question 33

The research assistant application includes Bedrock reranker model in its architecture. What is its primary role?

1. Improve recall when a user query is too narrow or uses alternate terminology
2. Split uniform documents into predictable token or character windows for retrieval
3. Perform nearest-neighbor search over embeddings with scalable search infrastructure
4. Reorder initially retrieved passages so the most contextually relevant items appear first

Correct Answer: D

 

Correct Answer

Answer D is correct because Bedrock reranker model is specifically used to reorder initially retrieved passages so the most contextually relevant items appear first. It applies a second-stage relevance model to candidate results before context is sent to the foundation model.

Incorrect Answers

Answer A is incorrect because that requirement aligns with query expansion, not Bedrock reranker model.

Answer B is incorrect because that requirement aligns with fixed-size chunking, not Bedrock reranker model.

Answer C is incorrect because that requirement aligns with OpenSearch semantic vector retrieval, not Bedrock reranker model.

 

Question 34

The AI product engineering team plans to adopt query expansion. Which outcome should drive that decision?

1. Perform vector similarity search alongside relational filtering and transactional data
2. Improve recall when a user query is too narrow or uses alternate terminology
3. Break a complex request into retrievable subquestions or normalize it for better search
4. Preserve sections and parent-child context when documents have meaningful structure

Correct Answer: B

 

Correct Answer

Answer B is correct because query expansion is specifically used to improve recall when a user query is too narrow or uses alternate terminology. It adds related terms or reformulations so retrieval can find relevant content that the original wording might miss.

Incorrect Answers

Answer A is incorrect because that requirement aligns with Aurora pgvector retrieval, not query expansion.

Answer C is incorrect because that requirement aligns with query decomposition and transformation, not query expansion.

Answer D is incorrect because that requirement aligns with hierarchical chunking, not query expansion.

 

Question 35

Within the supply-chain analytics group’s design, the team highlights query decomposition and transformation. Which need does that component address?

1. Reduce custom retrieval plumbing for a standard RAG application
2. Let agents and foundation models access retrieval tools through a consistent contract
3. Break a complex request into retrievable subquestions or normalize it for better search
4. Convert text into vector representations for semantic retrieval

Correct Answer: C

 

Correct Answer

Answer C is correct because query decomposition and transformation is specifically used to break a complex request into retrievable subquestions or normalize it for better search. It uses logic or models to rewrite, split, or sequence retrieval queries before collecting context.

Incorrect Answers

Answer A is incorrect because that requirement aligns with Bedrock Knowledge Bases managed retrieval, not query decomposition and transformation.

Answer B is incorrect because that requirement aligns with standardized retrieval interface with MCP or function calling, not query decomposition and transformation.

Answer D is incorrect because that requirement aligns with Amazon Titan embedding model, not query decomposition and transformation.