Amazon AWS AI Practitioner AIF-C01 Vector Stores Customization Tradeoffs And AI Agents Practice Test

 

AIF-C01 skills 3.1 | 30 original questions

This AWS Certified AI Practitioner AIF-C01 practice test focuses on vector stores customization tradeoffs and ai agents through original scenario-based questions aligned to AWS Exam Guide version 1.1 published April 30, 2026. Use the full ExamSnap AIF-C01 collection for broader practice across all five current exam domains. For broader exam preparation, review the Amazon AWS Certified AI Practitioner AIF-C01 Exam Dumps page.

Instructions: Select the best answer for each question. Review the rationale after answering. Each distractor includes a brief explanation of why it is not the strongest fit for the stated scenario.

Question 1

For the fraud-review pilot at Alpine Ski House, stakeholders need to decompose a question, search approved sources, compare evidence, and synthesize a result. Which concept, service, or technique most directly addresses this goal? Operational ownership is already assigned, so the team is comparing technical fit rather than staffing models. The control owner requires evidence from 5 test groups before the 395-day release review.

1. Multi-agent pattern
2. Workflow orchestration
3. Tool use
4. Multi-agent communication
5. Research agent

Correct answer: E

Why: Agents can coordinate iterative retrieval and reasoning tasks. It directly addresses the requirement in this scenario.

Option review:

A: Multi-agent systems use multiple specialized agents to collaborate on complex tasks. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

B: Orchestration governs the sequence and control flow of multi-step agent behavior. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

C: Tools let agents retrieve fresh data or perform actions beyond text generation. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

D: Communication patterns allow agents to coordinate work and share intermediate outcomes. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

E: Agents can coordinate iterative retrieval and reasoning tasks. It directly addresses the requirement in this scenario.

Learning point: Research agent – Agents can coordinate iterative retrieval and reasoning tasks.

Question 2

Humongous Insurance is comparing alternatives for its analytics modernization. The security architect needs to combine graph relationships with vector-based similarity capabilities for connected-data applications. Which option is most appropriate while trying to use current managed AWS capabilities? Existing application interfaces can accommodate any of the listed choices, so functional fit is the deciding factor. The project has 2 downstream consumers and a monthly review of approximately 432 sampled interactions.

1. Amazon Neptune
2. Neural network
3. Natural language processing (NLP)
4. Artificial intelligence (AI)
5. Computer vision

Correct answer: A

Why: Amazon Neptune is an in-scope option for vector storage/search where graph context is relevant. It directly addresses the requirement in this scenario.

Option review:

A: Amazon Neptune is an in-scope option for vector storage/search where graph context is relevant. It directly addresses the requirement in this scenario.

B: A neural network transforms inputs through connected layers whose parameters are learned during training. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

C: NLP covers techniques for understanding, extracting information from, and generating human language. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

D: AI is the umbrella field that includes many approaches such as machine learning, reasoning, perception, and language processing. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

E: Computer vision applies AI/ML to visual data such as images and video. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

Learning point: Amazon Neptune – Amazon Neptune is an in-scope option for vector storage/search where graph context is relevant.

Question 3

An architecture review at Graphic Design Institute has narrowed a compliance-assistant prototype decision to one requirement: change model weights using curated examples when repeatable domain behavior or style is required. What should the risk manager select? Assume the required AWS capabilities are available in the selected Region and normal governance controls are in place. The rollout spans 7 application teams, each using the same approved requirement set for the next 469 days.

1. Continuous pre-training
2. Fine-tuning
3. Retrieval Augmented Generation (RAG)
4. Model distillation
5. Pre-training from scratch

Correct answer: B

Why: Fine-tuning adds training cost but can improve task-specific behavior without full pre-training. It directly addresses the requirement in this scenario.

Option review:

A: Continuous pre-training is useful when substantial domain language or knowledge must be incorporated. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

B: Fine-tuning adds training cost but can improve task-specific behavior without full pre-training. It directly addresses the requirement in this scenario.

C: RAG is attractive for changing factual knowledge and source attribution but adds retrieval infrastructure and latency. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

D: Distillation can retain useful behavior in a smaller model while lowering serving cost. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

E: Pre-training is the most resource-intensive customization path and is rarely justified for ordinary application adaptation. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

Learning point: Fine-tuning – Fine-tuning adds training cost but can improve task-specific behavior without full pre-training.

Question 4

The security architect at Relecloud is preparing a recommendation for a forecasting initiative. The recommendation must plan and execute multiple steps toward a goal using a model, tools, and state. Which choice is the best match? The review committee wants a direct mapping from the requirement to the chosen capability. The evaluation set contains examples from 4 business workflows and 506 recent production cases.

1. Tool-using support agent
2. Multi-agent communication
3. AI agent
4. Workflow orchestration
5. Agent memory

Correct answer: C

Why: An AI agent extends model generation with goal-directed action and orchestration. It directly addresses the requirement in this scenario.

Option review:

A: Agents are appropriate when the workflow requires both reasoning and actions in external systems. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

B: Communication patterns allow agents to coordinate work and share intermediate outcomes. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

C: An AI agent extends model generation with goal-directed action and orchestration. It directly addresses the requirement in this scenario.

D: Orchestration governs the sequence and control flow of multi-step agent behavior. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

E: Memory helps agents maintain continuity beyond a single model invocation. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

Learning point: AI agent – An AI agent extends model generation with goal-directed action and orchestration.

Question 5

Adventure Works Manufacturing has completed discovery for a customer-support modernization. Before implementation, the risk manager must decide how to store embeddings and perform vector similarity search in a managed search platform. Which choice best satisfies that requirement? The solution will serve multiple internal teams, so the recommendation should be reusable without changing the core requirement. The initial rollout covers 543 internal users across 9 business units.

1. Machine learning (ML)
2. Amazon Neptune
3. Amazon Aurora PostgreSQL-Compatible Edition
4. Amazon OpenSearch Service
5. Amazon RDS for PostgreSQL

Correct answer: D

Why: OpenSearch Service supports vector search and is commonly used for semantic retrieval. It directly addresses the requirement in this scenario.

Option review:

A: ML is a subset of AI in which algorithms learn relationships from data to make predictions or decisions. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

B: Amazon Neptune is an in-scope option for vector storage/search where graph context is relevant. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

C: Aurora PostgreSQL can support vector workloads alongside relational application data. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

D: OpenSearch Service supports vector search and is commonly used for semantic retrieval. It directly addresses the requirement in this scenario.

E: RDS for PostgreSQL can host vector representations when a PostgreSQL operational model fits the application. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

Learning point: Amazon OpenSearch Service – OpenSearch Service supports vector search and is commonly used for semantic retrieval.

Question 6

While planning a agentic workflow trial, Proseware Services identifies this requirement: build a new foundation model when ownership and unique capability justify the largest data, compute, and engineering investment. Which option should the security architect prioritize if the goal is to keep the design easy to explain? The decision must follow the workload characteristics rather than a preference for the largest model or newest service. The workload processes about 580 requests during its busiest hour and has a documented fallback path.

1. Transfer learning
2. Domain adaptation
3. Retrieval Augmented Generation (RAG)
4. Pre-training
5. Pre-training from scratch

Correct answer: E

Why: Pre-training is the most resource-intensive customization path and is rarely justified for ordinary application adaptation. It directly addresses the requirement in this scenario.

Option review:

A: Transfer learning reduces the need to learn everything from scratch. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

B: Domain adaptation specializes a general model for a target field. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

C: RAG is attractive for changing factual knowledge and source attribution but adds retrieval infrastructure and latency. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

D: Pre-training establishes the base capabilities of a foundation model. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

E: Pre-training is the most resource-intensive customization path and is rarely justified for ordinary application adaptation. It directly addresses the requirement in this scenario.

Learning point: Pre-training from scratch – Pre-training is the most resource-intensive customization path and is rarely justified for ordinary application adaptation.

Question 7

A proof of concept at Lucerne Publishing exposed a design decision for the risk manager: the solution must monitor an approved signal, gather context, propose remediation, and request human approval before a sensitive action. Which option most directly solves that problem? The security baseline is already defined; the decision here concerns the specific capability described in the requirement. The pilot uses 617 representative records from 3 approved data sources.

1. Operations agent
2. Workflow orchestration
3. Multi-agent pattern
4. Research agent
5. Tool use

Correct answer: A

Why: Agents can orchestrate multi-step operational workflows with explicit guardrails and approval points. It directly addresses the requirement in this scenario.

Option review:

A: Agents can orchestrate multi-step operational workflows with explicit guardrails and approval points. It directly addresses the requirement in this scenario.

B: Orchestration governs the sequence and control flow of multi-step agent behavior. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

C: Multi-agent systems use multiple specialized agents to collaborate on complex tasks. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

D: Agents can coordinate iterative retrieval and reasoning tasks. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

E: Tools let agents retrieve fresh data or perform actions beyond text generation. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

Learning point: Operations agent – Agents can orchestrate multi-step operational workflows with explicit guardrails and approval points.

Question 8

Lamna Healthcare is documenting the target state for a operations automation program. The security architect needs a solution that can store embeddings in a managed PostgreSQL database using supported vector extensions. Which option is the strongest fit? The recommendation must solve the stated requirement without introducing unrelated platform complexity. The first release supports 8 departments and is reviewed every 654 days.

1. Amazon Neptune
2. Amazon RDS for PostgreSQL
3. Inference
4. Computer vision
5. Bias

Correct answer: B

Why: RDS for PostgreSQL can host vector representations when a PostgreSQL operational model fits the application. It directly addresses the requirement in this scenario.

Option review:

A: Amazon Neptune is an in-scope option for vector storage/search where graph context is relevant. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

B: RDS for PostgreSQL can host vector representations when a PostgreSQL operational model fits the application. It directly addresses the requirement in this scenario.

C: Inference is the execution phase in which a trained model processes new data. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

D: Computer vision applies AI/ML to visual data such as images and video. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

E: Bias is systematic error or skew that can affect predictions and fairness. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

Learning point: Amazon RDS for PostgreSQL – RDS for PostgreSQL can host vector representations when a PostgreSQL operational model fits the application.

Question 9

Contoso Retail is reviewing a sales-assistant rollout. The risk manager has one primary requirement: inject fresh enterprise knowledge at inference time rather than encoding that knowledge into model weights. Which choice best fits the requirement? The design must remain supportable after launch, but no additional feature is required beyond the stated need. The service has a 691-millisecond internal response target for the affected workflow.

1. Continuous pre-training
2. Distillation
3. Retrieval Augmented Generation (RAG)
4. Transfer learning
5. Pre-training from scratch

Correct answer: C

Why: RAG is attractive for changing factual knowledge and source attribution but adds retrieval infrastructure and latency. It directly addresses the requirement in this scenario.

Option review:

A: Continuous pre-training is useful when substantial domain language or knowledge must be incorporated. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

B: Distillation aims to reduce serving cost and latency while retaining much of the teacher capability. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

C: RAG is attractive for changing factual knowledge and source attribution but adds retrieval infrastructure and latency. It directly addresses the requirement in this scenario.

D: Transfer learning reduces the need to learn everything from scratch. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

E: Pre-training is the most resource-intensive customization path and is rarely justified for ordinary application adaptation. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

Learning point: Retrieval Augmented Generation (RAG) – RAG is attractive for changing factual knowledge and source attribution but adds retrieval infrastructure and latency.

Question 10

During a design review for Fourth Coffee, the security architect must look up an order, update a ticket, and draft a response under defined permissions. The team also wants to control recurring cost. What should the team choose? A short pilot window means the team prefers an approach that can be evaluated with clear success criteria. The team is comparing 2 candidate designs after a 728-day proof of concept.

1. AI agent
2. Model Context Protocol (MCP)
3. Research agent
4. Tool-using support agent
5. Operations agent

Correct answer: D

Why: Agents are appropriate when the workflow requires both reasoning and actions in external systems. It directly addresses the requirement in this scenario.

Option review:

A: An AI agent extends model generation with goal-directed action and orchestration. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

B: MCP provides a standard way for AI systems to discover and use external capabilities. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

C: Agents can coordinate iterative retrieval and reasoning tasks. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

D: Agents are appropriate when the workflow requires both reasoning and actions in external systems. It directly addresses the requirement in this scenario.

E: Agents can orchestrate multi-step operational workflows with explicit guardrails and approval points. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

Learning point: Tool-using support agent – Agents are appropriate when the workflow requires both reasoning and actions in external systems.

Question 11

Margie Travel is moving a document-intelligence project from pilot to production. The key decision is how to store relational data and vector embeddings with PostgreSQL vector extensions in a managed Aurora database. Which option is the strongest fit if the team wants to meet a strict latency target? The architecture board will reject a choice that addresses a different problem from the one described. The control owner requires evidence from 7 test groups before the 765-day release review.

1. Artificial intelligence (AI)
2. Amazon OpenSearch Service
3. Inference
4. Deep learning
5. Amazon Aurora PostgreSQL-Compatible Edition

Correct answer: E

Why: Aurora PostgreSQL can support vector workloads alongside relational application data. It directly addresses the requirement in this scenario.

Option review:

A: AI is the umbrella field that includes many approaches such as machine learning, reasoning, perception, and language processing. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

B: OpenSearch Service supports vector search and is commonly used for semantic retrieval. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

C: Inference is the execution phase in which a trained model processes new data. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

D: Deep learning is an ML approach based on neural networks with many layers and is commonly used for complex image, speech, and language tasks. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

E: Aurora PostgreSQL can support vector workloads alongside relational application data. It directly addresses the requirement in this scenario.

Learning point: Amazon Aurora PostgreSQL-Compatible Edition – Aurora PostgreSQL can support vector workloads alongside relational application data.

Question 12

A workshop at School of Fine Art focuses on a single decision: how to show examples or instructions in the prompt when adaptation can be achieved without changing model weights. Which option should the security architect recommend? Budget has been approved for the project, but the team still wants to avoid unnecessary recurring consumption. The project has 4 downstream consumers and a monthly review of approximately 802 sampled interactions.

1. In-context learning
2. Retrieval Augmented Generation (RAG)
3. Distillation
4. Continuous pre-training
5. Fine-tuning

Correct answer: A

Why: In-context learning is fast to iterate but consumes context tokens on each request. It directly addresses the requirement in this scenario.

Option review:

A: In-context learning is fast to iterate but consumes context tokens on each request. It directly addresses the requirement in this scenario.

B: RAG is attractive for changing factual knowledge and source attribution but adds retrieval infrastructure and latency. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

C: Distillation aims to reduce serving cost and latency while retaining much of the teacher capability. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

D: Continuous pre-training is useful when substantial domain language or knowledge must be incorporated. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

E: Fine-tuning adds training cost but can improve task-specific behavior without full pre-training. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

Learning point: In-context learning – In-context learning is fast to iterate but consumes context tokens on each request.

Question 13

For the claims-processing redesign at Northwind Analytics, stakeholders need to decompose a question, search approved sources, compare evidence, and synthesize a result. Which concept, service, or technique most directly addresses this goal? The team will validate the result with representative production examples before rollout. The rollout spans 9 application teams, each using the same approved requirement set for the next 839 days.

1. AI agent
2. Research agent
3. Model Context Protocol (MCP)
4. Multi-agent pattern
5. Tool use

Correct answer: B

Why: Agents can coordinate iterative retrieval and reasoning tasks. It directly addresses the requirement in this scenario.

Option review:

A: An AI agent extends model generation with goal-directed action and orchestration. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

B: Agents can coordinate iterative retrieval and reasoning tasks. It directly addresses the requirement in this scenario.

C: MCP provides a standard way for AI systems to discover and use external capabilities. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

D: Multi-agent systems use multiple specialized agents to collaborate on complex tasks. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

E: Tools let agents retrieve fresh data or perform actions beyond text generation. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

Learning point: Research agent – Agents can coordinate iterative retrieval and reasoning tasks.

Question 14

Litware Financial is comparing alternatives for its personalization program. The security architect needs to combine graph relationships with vector-based similarity capabilities for connected-data applications. Which option is most appropriate while trying to use current managed AWS capabilities? The pilot has representative data, and the team will measure the selected approach against an agreed acceptance threshold. The evaluation set contains examples from 6 business workflows and 876 recent production cases.

1. Machine learning (ML)
2. Amazon Aurora PostgreSQL-Compatible Edition
3. Amazon Neptune
4. Bias
5. Computer vision

Correct answer: C

Why: Amazon Neptune is an in-scope option for vector storage/search where graph context is relevant. It directly addresses the requirement in this scenario.

Option review:

A: ML is a subset of AI in which algorithms learn relationships from data to make predictions or decisions. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

B: Aurora PostgreSQL can support vector workloads alongside relational application data. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

C: Amazon Neptune is an in-scope option for vector storage/search where graph context is relevant. It directly addresses the requirement in this scenario.

D: Bias is systematic error or skew that can affect predictions and fairness. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

E: Computer vision applies AI/ML to visual data such as images and video. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

Learning point: Amazon Neptune – Amazon Neptune is an in-scope option for vector storage/search where graph context is relevant.

Question 15

An architecture review at A. Datum Research has narrowed a developer-productivity pilot decision to one requirement: train a smaller student model from a larger teacher when reduced inference cost or latency justifies an additional training step. What should the risk manager select? The team will document the rationale for auditors and wants the recommendation to be defensible from the scenario facts. The initial rollout covers 913 internal users across 3 business units.

1. Fine-tuning
2. Domain adaptation
3. In-context learning
4. Model distillation
5. Continuous pre-training

Correct answer: D

Why: Distillation can retain useful behavior in a smaller model while lowering serving cost. It directly addresses the requirement in this scenario.

Option review:

A: Fine-tuning adds training cost but can improve task-specific behavior without full pre-training. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

B: Domain adaptation specializes a general model for a target field. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

C: In-context learning is fast to iterate but consumes context tokens on each request. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

D: Distillation can retain useful behavior in a smaller model while lowering serving cost. It directly addresses the requirement in this scenario.

E: Continuous pre-training is useful when substantial domain language or knowledge must be incorporated. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

Learning point: Model distillation – Distillation can retain useful behavior in a smaller model while lowering serving cost.

Question 16

The security architect at Coho Winery is preparing a recommendation for a fraud-review pilot. The recommendation must look up an order, update a ticket, and draft a response under defined permissions. Which choice is the best match? The team wants the least complex technically correct choice that satisfies the requirement. The workload processes about 950 requests during its busiest hour and has a documented fallback path.

1. Model Context Protocol (MCP)
2. Tool use
3. AI agent
4. Workflow orchestration
5. Tool-using support agent

Correct answer: E

Why: Agents are appropriate when the workflow requires both reasoning and actions in external systems. It directly addresses the requirement in this scenario.

Option review:

A: MCP provides a standard way for AI systems to discover and use external capabilities. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

B: Tools let agents retrieve fresh data or perform actions beyond text generation. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

C: An AI agent extends model generation with goal-directed action and orchestration. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

D: Orchestration governs the sequence and control flow of multi-step agent behavior. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

E: Agents are appropriate when the workflow requires both reasoning and actions in external systems. It directly addresses the requirement in this scenario.

Learning point: Tool-using support agent – Agents are appropriate when the workflow requires both reasoning and actions in external systems.

Question 17

Lucerne Retail has completed discovery for a analytics modernization. Before implementation, the risk manager must decide how to store relational data and vector embeddings with PostgreSQL vector extensions in a managed Aurora database. Which choice best satisfies that requirement? The workload has passed basic feasibility checks, so the remaining question is which approach best matches the requirement. The pilot uses 987 representative records from 5 approved data sources.

1. Amazon Aurora PostgreSQL-Compatible Edition
2. Natural language processing (NLP)
3. Inference
4. Amazon RDS for PostgreSQL
5. Amazon OpenSearch Service

Correct answer: A

Why: Aurora PostgreSQL can support vector workloads alongside relational application data. It directly addresses the requirement in this scenario.

Option review:

A: Aurora PostgreSQL can support vector workloads alongside relational application data. It directly addresses the requirement in this scenario.

B: NLP covers techniques for understanding, extracting information from, and generating human language. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

C: Inference is the execution phase in which a trained model processes new data. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

D: RDS for PostgreSQL can host vector representations when a PostgreSQL operational model fits the application. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

E: OpenSearch Service supports vector search and is commonly used for semantic retrieval. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

Learning point: Amazon Aurora PostgreSQL-Compatible Edition – Aurora PostgreSQL can support vector workloads alongside relational application data.

Question 18

While planning a compliance-assistant prototype, Tailspin Toys identifies this requirement: change model weights using curated examples when repeatable domain behavior or style is required. Which option should the security architect prioritize if the goal is to keep the design easy to explain? Stakeholders have ruled out a broad redesign and want the choice that most precisely addresses the stated need. The first release supports 2 departments and is reviewed every 64 days.

1. Domain adaptation
2. Fine-tuning
3. In-context learning
4. Model distillation
5. Continuous pre-training

Correct answer: B

Why: Fine-tuning adds training cost but can improve task-specific behavior without full pre-training. It directly addresses the requirement in this scenario.

Option review:

A: Domain adaptation specializes a general model for a target field. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

B: Fine-tuning adds training cost but can improve task-specific behavior without full pre-training. It directly addresses the requirement in this scenario.

C: In-context learning is fast to iterate but consumes context tokens on each request. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

D: Distillation can retain useful behavior in a smaller model while lowering serving cost. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

E: Continuous pre-training is useful when substantial domain language or knowledge must be incorporated. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

Learning point: Fine-tuning – Fine-tuning adds training cost but can improve task-specific behavior without full pre-training.

Question 19

A proof of concept at City Power and Light exposed a design decision for the risk manager: the solution must monitor an approved signal, gather context, propose remediation, and request human approval before a sensitive action. Which option most directly solves that problem? Operational ownership is already assigned, so the team is comparing technical fit rather than staffing models. The service has a 101-millisecond internal response target for the affected workflow.

1. Model Context Protocol (MCP)
2. Tool use
3. Operations agent
4. Agent memory
5. Multi-agent pattern

Correct answer: C

Why: Agents can orchestrate multi-step operational workflows with explicit guardrails and approval points. It directly addresses the requirement in this scenario.

Option review:

A: MCP provides a standard way for AI systems to discover and use external capabilities. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

B: Tools let agents retrieve fresh data or perform actions beyond text generation. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

C: Agents can orchestrate multi-step operational workflows with explicit guardrails and approval points. It directly addresses the requirement in this scenario.

D: Memory helps agents maintain continuity beyond a single model invocation. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

E: Multi-agent systems use multiple specialized agents to collaborate on complex tasks. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

Learning point: Operations agent – Agents can orchestrate multi-step operational workflows with explicit guardrails and approval points.

Question 20

Consolidated Messenger is documenting the target state for a customer-support modernization. The security architect needs a solution that can store embeddings in a managed PostgreSQL database using supported vector extensions. Which option is the strongest fit? Existing application interfaces can accommodate any of the listed choices, so functional fit is the deciding factor. The team is comparing 4 candidate designs after a 138-day proof of concept.

1. Bias
2. Amazon Neptune
3. Natural language processing (NLP)
4. Amazon RDS for PostgreSQL
5. Amazon OpenSearch Service

Correct answer: D

Why: RDS for PostgreSQL can host vector representations when a PostgreSQL operational model fits the application. It directly addresses the requirement in this scenario.

Option review:

A: Bias is systematic error or skew that can affect predictions and fairness. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

B: Amazon Neptune is an in-scope option for vector storage/search where graph context is relevant. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

C: NLP covers techniques for understanding, extracting information from, and generating human language. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

D: RDS for PostgreSQL can host vector representations when a PostgreSQL operational model fits the application. It directly addresses the requirement in this scenario.

E: OpenSearch Service supports vector search and is commonly used for semantic retrieval. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

Learning point: Amazon RDS for PostgreSQL – RDS for PostgreSQL can host vector representations when a PostgreSQL operational model fits the application.

Question 21

Nod Publishers is reviewing a agentic workflow trial. The risk manager has one primary requirement: show examples or instructions in the prompt when adaptation can be achieved without changing model weights. Which choice best fits the requirement? Assume the required AWS capabilities are available in the selected Region and normal governance controls are in place. The control owner requires evidence from 9 test groups before the 175-day release review.

1. Domain adaptation
2. Instruction tuning
3. Transfer learning
4. Model distillation
5. In-context learning

Correct answer: E

Why: In-context learning is fast to iterate but consumes context tokens on each request. It directly addresses the requirement in this scenario.

Option review:

A: Domain adaptation specializes a general model for a target field. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

B: Instruction tuning teaches a model to respond appropriately to explicit instructions. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

C: Transfer learning reduces the need to learn everything from scratch. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

D: Distillation can retain useful behavior in a smaller model while lowering serving cost. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

E: In-context learning is fast to iterate but consumes context tokens on each request. It directly addresses the requirement in this scenario.

Learning point: In-context learning – In-context learning is fast to iterate but consumes context tokens on each request.

Question 22

During a design review for Fabrikam Health, the security architect must plan and execute multiple steps toward a goal using a model, tools, and state. The team also wants to control recurring cost. What should the team choose? The review committee wants a direct mapping from the requirement to the chosen capability. The project has 6 downstream consumers and a monthly review of approximately 212 sampled interactions.

1. AI agent
2. Workflow orchestration
3. Operations agent
4. Tool use
5. Multi-agent communication

Correct answer: A

Why: An AI agent extends model generation with goal-directed action and orchestration. It directly addresses the requirement in this scenario.

Option review:

A: An AI agent extends model generation with goal-directed action and orchestration. It directly addresses the requirement in this scenario.

B: Orchestration governs the sequence and control flow of multi-step agent behavior. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

C: Agents can orchestrate multi-step operational workflows with explicit guardrails and approval points. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

D: Tools let agents retrieve fresh data or perform actions beyond text generation. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

E: Communication patterns allow agents to coordinate work and share intermediate outcomes. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

Learning point: AI agent – An AI agent extends model generation with goal-directed action and orchestration.

Question 23

Wingtip Logistics is moving a operations automation program from pilot to production. The key decision is how to store embeddings and perform vector similarity search in a managed search platform. Which option is the strongest fit if the team wants to meet a strict latency target? The solution will serve multiple internal teams, so the recommendation should be reusable without changing the core requirement. The rollout spans 3 application teams, each using the same approved requirement set for the next 249 days.

1. Neural network
2. Amazon OpenSearch Service
3. Amazon RDS for PostgreSQL
4. Machine learning (ML)
5. Bias

Correct answer: B

Why: OpenSearch Service supports vector search and is commonly used for semantic retrieval. It directly addresses the requirement in this scenario.

Option review:

A: A neural network transforms inputs through connected layers whose parameters are learned during training. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

B: OpenSearch Service supports vector search and is commonly used for semantic retrieval. It directly addresses the requirement in this scenario.

C: RDS for PostgreSQL can host vector representations when a PostgreSQL operational model fits the application. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

D: ML is a subset of AI in which algorithms learn relationships from data to make predictions or decisions. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

E: Bias is systematic error or skew that can affect predictions and fairness. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

Learning point: Amazon OpenSearch Service – OpenSearch Service supports vector search and is commonly used for semantic retrieval.

Question 24

A workshop at Trey Research focuses on a single decision: how to build a new foundation model when ownership and unique capability justify the largest data, compute, and engineering investment. Which option should the security architect recommend? The decision must follow the workload characteristics rather than a preference for the largest model or newest service. The evaluation set contains examples from 8 business workflows and 286 recent production cases.

1. Fine-tuning
2. Retrieval Augmented Generation (RAG)
3. Pre-training from scratch
4. Distillation
5. Continuous pre-training

Correct answer: C

Why: Pre-training is the most resource-intensive customization path and is rarely justified for ordinary application adaptation. It directly addresses the requirement in this scenario.

Option review:

A: Fine-tuning adds training cost but can improve task-specific behavior without full pre-training. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

B: RAG is attractive for changing factual knowledge and source attribution but adds retrieval infrastructure and latency. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

C: Pre-training is the most resource-intensive customization path and is rarely justified for ordinary application adaptation. It directly addresses the requirement in this scenario.

D: Distillation aims to reduce serving cost and latency while retaining much of the teacher capability. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

E: Continuous pre-training is useful when substantial domain language or knowledge must be incorporated. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

Learning point: Pre-training from scratch – Pre-training is the most resource-intensive customization path and is rarely justified for ordinary application adaptation.

Question 25

For the internal search upgrade at Bellows College, stakeholders need to decompose a question, search approved sources, compare evidence, and synthesize a result. Which concept, service, or technique most directly addresses this goal? The security baseline is already defined; the decision here concerns the specific capability described in the requirement. The initial rollout covers 323 internal users across 5 business units.

1. Multi-agent pattern
2. Multi-agent communication
3. Tool use
4. Research agent
5. Operations agent

Correct answer: D

Why: Agents can coordinate iterative retrieval and reasoning tasks. It directly addresses the requirement in this scenario.

Option review:

A: Multi-agent systems use multiple specialized agents to collaborate on complex tasks. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

B: Communication patterns allow agents to coordinate work and share intermediate outcomes. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

C: Tools let agents retrieve fresh data or perform actions beyond text generation. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

D: Agents can coordinate iterative retrieval and reasoning tasks. It directly addresses the requirement in this scenario.

E: Agents can orchestrate multi-step operational workflows with explicit guardrails and approval points. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

Learning point: Research agent – Agents can coordinate iterative retrieval and reasoning tasks.

Question 26

Blue Yonder Airlines is comparing alternatives for its document-intelligence project. The security architect needs to combine graph relationships with vector-based similarity capabilities for connected-data applications. Which option is most appropriate while trying to use current managed AWS capabilities? The recommendation must solve the stated requirement without introducing unrelated platform complexity. The workload processes about 360 requests during its busiest hour and has a documented fallback path.

1. Training
2. Machine learning (ML)
3. Amazon OpenSearch Service
4. Artificial intelligence (AI)
5. Amazon Neptune

Correct answer: E

Why: Amazon Neptune is an in-scope option for vector storage/search where graph context is relevant. It directly addresses the requirement in this scenario.

Option review:

A: Training is the process that fits model parameters using data and an optimization procedure. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

B: ML is a subset of AI in which algorithms learn relationships from data to make predictions or decisions. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

C: OpenSearch Service supports vector search and is commonly used for semantic retrieval. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

D: AI is the umbrella field that includes many approaches such as machine learning, reasoning, perception, and language processing. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

E: Amazon Neptune is an in-scope option for vector storage/search where graph context is relevant. It directly addresses the requirement in this scenario.

Learning point: Amazon Neptune – Amazon Neptune is an in-scope option for vector storage/search where graph context is relevant.

Question 27

An architecture review at Woodgrove Bank has narrowed a knowledge-assistant rollout decision to one requirement: train a smaller student model from a larger teacher when reduced inference cost or latency justifies an additional training step. What should the risk manager select? The design must remain supportable after launch, but no additional feature is required beyond the stated need. The pilot uses 397 representative records from 7 approved data sources.

1. Model distillation
2. Transfer learning
3. In-context learning
4. Domain adaptation
5. Pre-training

Correct answer: A

Why: Distillation can retain useful behavior in a smaller model while lowering serving cost. It directly addresses the requirement in this scenario.

Option review:

A: Distillation can retain useful behavior in a smaller model while lowering serving cost. It directly addresses the requirement in this scenario.

B: Transfer learning reduces the need to learn everything from scratch. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

C: In-context learning is fast to iterate but consumes context tokens on each request. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

D: Domain adaptation specializes a general model for a target field. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

E: Pre-training establishes the base capabilities of a foundation model. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

Learning point: Model distillation – Distillation can retain useful behavior in a smaller model while lowering serving cost.

Question 28

The security architect at Wide World Importers is preparing a recommendation for a claims-processing redesign. The recommendation must monitor an approved signal, gather context, propose remediation, and request human approval before a sensitive action. Which choice is the best match? A short pilot window means the team prefers an approach that can be evaluated with clear success criteria. The first release supports 4 departments and is reviewed every 434 days.

1. Agent memory
2. Operations agent
3. Workflow orchestration
4. Tool-using support agent
5. Tool use

Correct answer: B

Why: Agents can orchestrate multi-step operational workflows with explicit guardrails and approval points. It directly addresses the requirement in this scenario.

Option review:

A: Memory helps agents maintain continuity beyond a single model invocation. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

B: Agents can orchestrate multi-step operational workflows with explicit guardrails and approval points. It directly addresses the requirement in this scenario.

C: Orchestration governs the sequence and control flow of multi-step agent behavior. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

D: Agents are appropriate when the workflow requires both reasoning and actions in external systems. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

E: Tools let agents retrieve fresh data or perform actions beyond text generation. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

Learning point: Operations agent – Agents can orchestrate multi-step operational workflows with explicit guardrails and approval points.

Question 29

VanArsdel Media has completed discovery for a personalization program. Before implementation, the risk manager must decide how to store embeddings in a managed PostgreSQL database using supported vector extensions. Which choice best satisfies that requirement? The architecture board will reject a choice that addresses a different problem from the one described. The service has a 471-millisecond internal response target for the affected workflow.

1. Neural network
2. Deep learning
3. Amazon RDS for PostgreSQL
4. Inference
5. Training

Correct answer: C

Why: RDS for PostgreSQL can host vector representations when a PostgreSQL operational model fits the application. It directly addresses the requirement in this scenario.

Option review:

A: A neural network transforms inputs through connected layers whose parameters are learned during training. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

B: Deep learning is an ML approach based on neural networks with many layers and is commonly used for complex image, speech, and language tasks. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

C: RDS for PostgreSQL can host vector representations when a PostgreSQL operational model fits the application. It directly addresses the requirement in this scenario.

D: Inference is the execution phase in which a trained model processes new data. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

E: Training is the process that fits model parameters using data and an optimization procedure. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

Learning point: Amazon RDS for PostgreSQL – RDS for PostgreSQL can host vector representations when a PostgreSQL operational model fits the application.

Question 30

While planning a developer-productivity pilot, Datum Dynamics identifies this requirement: inject fresh enterprise knowledge at inference time rather than encoding that knowledge into model weights. Which option should the security architect prioritize if the goal is to keep the design easy to explain? Budget has been approved for the project, but the team still wants to avoid unnecessary recurring consumption. The team is comparing 6 candidate designs after a 508-day proof of concept.

1. Pre-training
2. Domain adaptation
3. Transfer learning
4. Retrieval Augmented Generation (RAG)
5. Model distillation

Correct answer: D

Why: RAG is attractive for changing factual knowledge and source attribution but adds retrieval infrastructure and latency. It directly addresses the requirement in this scenario.

Option review:

A: Pre-training establishes the base capabilities of a foundation model. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

B: Domain adaptation specializes a general model for a target field. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

C: Transfer learning reduces the need to learn everything from scratch. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

D: RAG is attractive for changing factual knowledge and source attribution but adds retrieval infrastructure and latency. It directly addresses the requirement in this scenario.

E: Distillation can retain useful behavior in a smaller model while lowering serving cost. This can be appropriate in another scenario, but it does not most directly satisfy the requirement described here.

Learning point: Retrieval Augmented Generation (RAG) – RAG is attractive for changing factual knowledge and source attribution but adds retrieval infrastructure and latency.