Amazon AWS AIP-C01 Foundation Model Selection and Lifecycle Practice Test

 

Topic 02 focuses on Foundation Model Selection, Configuration, and Lifecycle 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 internal AI platform team is prioritizing a requirement to select a model by comparing quality, modality, context limits, latency, cost, and task-specific performance. Which implementation is most appropriate?

1. Amazon SageMaker AI endpoint for a customized model
2. foundation model benchmark and capability evaluation
3. automated model deployment with rollback
4. model abstraction layer with Lambda, API Gateway, and AppConfig

Correct Answer: B

 

Correct Answer

Answer B is correct because foundation model benchmark and capability evaluation is designed to select a model by comparing quality, modality, context limits, latency, cost, and task-specific performance. It uses representative prompts and measurable acceptance criteria rather than choosing a model only by size or popularity.

Incorrect Answers

Answer A is incorrect because Amazon SageMaker AI endpoint for a customized model is primarily used to serve a domain-specific or fine-tuned model that requires managed custom model hosting, which is a different requirement from the one being tested.

Answer C is incorrect because automated model deployment with rollback is primarily used to update customized models safely and recover quickly from a failed release, which is a different requirement from the one being tested.

Answer D is incorrect because model abstraction layer with Lambda, API Gateway, and AppConfig is primarily used to switch models or providers without rewriting application business logic, which is a different requirement from the one being tested.

 

Question 2

The data privacy office is prioritizing a requirement to switch models or providers without rewriting application business logic. Which implementation is most appropriate?

1. Amazon Bedrock Cross-Region Inference
2. model cascading
3. LoRA or adapter-based parameter-efficient adaptation
4. model abstraction layer with Lambda, API Gateway, and AppConfig

Correct Answer: D

 

Correct Answer

Answer D is correct because model abstraction layer with Lambda, API Gateway, and AppConfig is designed to switch models or providers without rewriting application business logic. It places model selection and configuration behind a stable application interface and externalized configuration.

Incorrect Answers

Answer A is incorrect because Amazon Bedrock Cross-Region Inference is primarily used to improve resilience or capacity access when a supported model has regional constraints, which is a different requirement from the one being tested.

Answer B is incorrect because model cascading is primarily used to use inexpensive or specialized models for routine requests while reserving larger models for harder requests, which is a different requirement from the one being tested.

Answer C is incorrect because LoRA or adapter-based parameter-efficient adaptation is primarily used to customize model behavior with fewer trainable parameters than full fine-tuning, which is a different requirement from the one being tested.

 

Question 3

The enterprise integration office is prioritizing a requirement to improve resilience or capacity access when a supported model has regional constraints. Which implementation is most appropriate?

1. Amazon SageMaker AI endpoint for a customized model
2. smaller specialized foundation model
3. Amazon Bedrock Cross-Region Inference
4. SageMaker Model Registry

Correct Answer: C

 

Correct Answer

Answer C is correct because Amazon Bedrock Cross-Region Inference is designed to improve resilience or capacity access when a supported model has regional constraints. It routes eligible Bedrock inference requests across supported regions according to a cross-region inference profile.

Incorrect Answers

Answer A is incorrect because Amazon SageMaker AI endpoint for a customized model is primarily used to serve a domain-specific or fine-tuned model that requires managed custom model hosting, which is a different requirement from the one being tested.

Answer B is incorrect because smaller specialized foundation model is primarily used to reduce latency and inference cost for a narrow task that does not need the capabilities of a larger model, which is a different requirement from the one being tested.

Answer D is incorrect because SageMaker Model Registry is primarily used to version and govern customized model artifacts across their deployment lifecycle, which is a different requirement from the one being tested.

 

Question 4

The corporate knowledge assistant needs an architecture that can serve a domain-specific or fine-tuned model that requires managed custom model hosting. Which choice best meets that need?

1. Amazon SageMaker AI endpoint for a customized model
2. automated model deployment with rollback
3. LoRA or adapter-based parameter-efficient adaptation
4. model inference parameter configuration

Correct Answer: A

 

Correct Answer

Answer A is correct because Amazon SageMaker AI endpoint for a customized model is designed to serve a domain-specific or fine-tuned model that requires managed custom model hosting. It deploys a controlled model artifact behind a scalable managed inference endpoint.

Incorrect Answers

Answer B is incorrect because automated model deployment with rollback is primarily used to update customized models safely and recover quickly from a failed release, which is a different requirement from the one being tested.

Answer C is incorrect because LoRA or adapter-based parameter-efficient adaptation is primarily used to customize model behavior with fewer trainable parameters than full fine-tuning, which is a different requirement from the one being tested.

Answer D is incorrect because model inference parameter configuration is primarily used to tune response determinism, diversity, and generation behavior for a specific use case, which is a different requirement from the one being tested.

 

Question 5

The enterprise data marketplace must customize model behavior with fewer trainable parameters than full fine-tuning. Which approach is the strongest fit?

1. SageMaker Model Registry
2. LoRA or adapter-based parameter-efficient adaptation
3. model cascading
4. foundation model benchmark and capability evaluation

Correct Answer: B

 

Correct Answer

Answer B is correct because LoRA or adapter-based parameter-efficient adaptation is designed to customize model behavior with fewer trainable parameters than full fine-tuning. It adds or updates compact adaptation parameters while leaving most base-model weights unchanged.

Incorrect Answers

Answer A is incorrect because SageMaker Model Registry is primarily used to version and govern customized model artifacts across their deployment lifecycle, which is a different requirement from the one being tested.

Answer C is incorrect because model cascading is primarily used to use inexpensive or specialized models for routine requests while reserving larger models for harder requests, which is a different requirement from the one being tested.

Answer D is incorrect because foundation model benchmark and capability evaluation is primarily used to select a model by comparing quality, modality, context limits, latency, cost, and task-specific performance, which is a different requirement from the one being tested.

 

Question 6

The case management platform must version and govern customized model artifacts across their deployment lifecycle. Which approach is the strongest fit?

1. SageMaker Model Registry
2. model abstraction layer with Lambda, API Gateway, and AppConfig
3. automated model deployment with rollback
4. smaller specialized foundation model

Correct Answer: A

 

Correct Answer

Answer A is correct because SageMaker Model Registry is designed to version and govern customized model artifacts across their deployment lifecycle. It tracks model versions, metadata, approval state, and promotion decisions for controlled deployment.

Incorrect Answers

Answer B is incorrect because model abstraction layer with Lambda, API Gateway, and AppConfig is primarily used to switch models or providers without rewriting application business logic, which is a different requirement from the one being tested.

Answer C is incorrect because automated model deployment with rollback is primarily used to update customized models safely and recover quickly from a failed release, which is a different requirement from the one being tested.

Answer D is incorrect because smaller specialized foundation model is primarily used to reduce latency and inference cost for a narrow task that does not need the capabilities of a larger model, which is a different requirement from the one being tested.

 

Question 7

The legal research platform is prioritizing a requirement to update customized models safely and recover quickly from a failed release. Which implementation is most appropriate?

1. model inference parameter configuration
2. model cascading
3. automated model deployment with rollback
4. Amazon Bedrock Cross-Region Inference

Correct Answer: C

 

Correct Answer

Answer C is correct because automated model deployment with rollback is designed to update customized models safely and recover quickly from a failed release. It uses versioned artifacts, deployment automation, health checks, and rollback criteria instead of replacing models manually.

Incorrect Answers

Answer A is incorrect because model inference parameter configuration is primarily used to tune response determinism, diversity, and generation behavior for a specific use case, which is a different requirement from the one being tested.

Answer B is incorrect because model cascading is primarily used to use inexpensive or specialized models for routine requests while reserving larger models for harder requests, which is a different requirement from the one being tested.

Answer D is incorrect because Amazon Bedrock Cross-Region Inference is primarily used to improve resilience or capacity access when a supported model has regional constraints, which is a different requirement from the one being tested.

 

Question 8

The research assistant application has a design goal to use inexpensive or specialized models for routine requests while reserving larger models for harder requests. What should the team choose?

1. smaller specialized foundation model
2. Amazon SageMaker AI endpoint for a customized model
3. model cascading
4. foundation model benchmark and capability evaluation

Correct Answer: C

 

Correct Answer

Answer C is correct because model cascading is designed to use inexpensive or specialized models for routine requests while reserving larger models for harder requests. It routes work through a sequence or hierarchy of models based on complexity, confidence, or policy.

Incorrect Answers

Answer A is incorrect because smaller specialized foundation model is primarily used to reduce latency and inference cost for a narrow task that does not need the capabilities of a larger model, which is a different requirement from the one being tested.

Answer B is incorrect because Amazon SageMaker AI endpoint for a customized model is primarily used to serve a domain-specific or fine-tuned model that requires managed custom model hosting, which is a different requirement from the one being tested.

Answer D is incorrect because foundation model benchmark and capability evaluation is primarily used to select a model by comparing quality, modality, context limits, latency, cost, and task-specific performance, which is a different requirement from the one being tested.

 

Question 9

The AI product engineering team needs an architecture that can reduce latency and inference cost for a narrow task that does not need the capabilities of a larger model. Which choice best meets that need?

1. smaller specialized foundation model
2. LoRA or adapter-based parameter-efficient adaptation
3. model abstraction layer with Lambda, API Gateway, and AppConfig
4. model inference parameter configuration

Correct Answer: A

 

Correct Answer

Answer A is correct because smaller specialized foundation model is designed to reduce latency and inference cost for a narrow task that does not need the capabilities of a larger model. It selects the smallest model that still satisfies measured quality and functional requirements.

Incorrect Answers

Answer B is incorrect because LoRA or adapter-based parameter-efficient adaptation is primarily used to customize model behavior with fewer trainable parameters than full fine-tuning, which is a different requirement from the one being tested.

Answer C is incorrect because model abstraction layer with Lambda, API Gateway, and AppConfig is primarily used to switch models or providers without rewriting application business logic, which is a different requirement from the one being tested.

Answer D is incorrect because model inference parameter configuration is primarily used to tune response determinism, diversity, and generation behavior for a specific use case, which is a different requirement from the one being tested.

 

Question 10

The supply-chain analytics group is prioritizing a requirement to tune response determinism, diversity, and generation behavior for a specific use case. Which implementation is most appropriate?

1. foundation model benchmark and capability evaluation
2. SageMaker Model Registry
3. model inference parameter configuration
4. Amazon Bedrock Cross-Region Inference

Correct Answer: C

 

Correct Answer

Answer C is correct because model inference parameter configuration is designed to tune response determinism, diversity, and generation behavior for a specific use case. It adjusts supported inference parameters such as temperature, top-p, top-k, stop conditions, and maximum output length.

Incorrect Answers

Answer A is incorrect because foundation model benchmark and capability evaluation is primarily used to select a model by comparing quality, modality, context limits, latency, cost, and task-specific performance, which is a different requirement from the one being tested.

Answer B is incorrect because SageMaker Model Registry is primarily used to version and govern customized model artifacts across their deployment lifecycle, which is a different requirement from the one being tested.

Answer D is incorrect because Amazon Bedrock Cross-Region Inference is primarily used to improve resilience or capacity access when a supported model has regional constraints, which is a different requirement from the one being tested.

 

Question 11

The knowledge management program describes a component that uses representative prompts and measurable acceptance criteria rather than choosing a model only by size or popularity. Which capability is being described?

1. Amazon SageMaker AI endpoint for a customized model
2. model abstraction layer with Lambda, API Gateway, and AppConfig
3. automated model deployment with rollback
4. foundation model benchmark and capability evaluation

Correct Answer: D

 

Correct Answer

Answer D is correct because the description directly matches foundation model benchmark and capability evaluation. foundation model benchmark and capability evaluation uses representative prompts and measurable acceptance criteria rather than choosing a model only by size or popularity.

Incorrect Answers

Answer A is incorrect because Amazon SageMaker AI endpoint for a customized model is primarily used to serve a domain-specific or fine-tuned model that requires managed custom model hosting, which is a different requirement from the one being tested.

Answer B is incorrect because model abstraction layer with Lambda, API Gateway, and AppConfig is primarily used to switch models or providers without rewriting application business logic, which is a different requirement from the one being tested.

Answer C is incorrect because automated model deployment with rollback is primarily used to update customized models safely and recover quickly from a failed release, which is a different requirement from the one being tested.

 

Question 12

The digital commerce architecture team needs to identify a capability with this behavior: places model selection and configuration behind a stable application interface and externalized configuration. What is the best match?

1. model abstraction layer with Lambda, API Gateway, and AppConfig
2. model cascading
3. LoRA or adapter-based parameter-efficient adaptation
4. Amazon Bedrock Cross-Region Inference

Correct Answer: A

 

Correct Answer

Answer A is correct because the description directly matches model abstraction layer with Lambda, API Gateway, and AppConfig. model abstraction layer with Lambda, API Gateway, and AppConfig places model selection and configuration behind a stable application interface and externalized configuration.

Incorrect Answers

Answer B is incorrect because model cascading is primarily used to use inexpensive or specialized models for routine requests while reserving larger models for harder requests, which is a different requirement from the one being tested.

Answer C is incorrect because LoRA or adapter-based parameter-efficient adaptation is primarily used to customize model behavior with fewer trainable parameters than full fine-tuning, which is a different requirement from the one being tested.

Answer D is incorrect because Amazon Bedrock Cross-Region Inference is primarily used to improve resilience or capacity access when a supported model has regional constraints, which is a different requirement from the one being tested.

 

Question 13

The AI reliability engineering team documents this GenAI behavior: routes eligible Bedrock inference requests across supported Regions according to a cross-Region inference profile. Which capability matches it?

1. Amazon Bedrock Cross-Region Inference
2. SageMaker Model Registry
3. smaller specialized foundation model
4. Amazon SageMaker AI endpoint for a customized model

Correct Answer: A

 

Correct Answer

Answer A is correct because the description directly matches Amazon Bedrock Cross-Region Inference. Amazon Bedrock Cross-Region Inference routes eligible Bedrock inference requests across supported Regions according to a cross-Region inference profile.

Incorrect Answers

Answer B is incorrect because SageMaker Model Registry is primarily used to version and govern customized model artifacts across their deployment lifecycle, which is a different requirement from the one being tested.

Answer C is incorrect because smaller specialized foundation model is primarily used to reduce latency and inference cost for a narrow task that does not need the capabilities of a larger model, which is a different requirement from the one being tested.

Answer D is incorrect because Amazon SageMaker AI endpoint for a customized model is primarily used to serve a domain-specific or fine-tuned model that requires managed custom model hosting, which is a different requirement from the one being tested.

 

Question 14

The field analytics service needs to identify a capability with this behavior: deploys a controlled model artifact behind a scalable managed inference endpoint. What is the best match?

1. LoRA or adapter-based parameter-efficient adaptation
2. automated model deployment with rollback
3. model inference parameter configuration
4. Amazon SageMaker AI endpoint for a customized model

Correct Answer: D

 

Correct Answer

Answer D is correct because the description directly matches Amazon SageMaker AI endpoint for a customized model. Amazon SageMaker AI endpoint for a customized model deploys a controlled model artifact behind a scalable managed inference endpoint.

Incorrect Answers

Answer A is incorrect because LoRA or adapter-based parameter-efficient adaptation is primarily used to customize model behavior with fewer trainable parameters than full fine-tuning, which is a different requirement from the one being tested.

Answer B is incorrect because automated model deployment with rollback is primarily used to update customized models safely and recover quickly from a failed release, which is a different requirement from the one being tested.

Answer C is incorrect because model inference parameter configuration is primarily used to tune response determinism, diversity, and generation behavior for a specific use case, which is a different requirement from the one being tested.

 

Question 15

The operations intelligence team needs to identify a capability with this behavior: adds or updates compact adaptation parameters while leaving most base-model weights unchanged. What is the best match?

1. model cascading
2. foundation model benchmark and capability evaluation
3. LoRA or adapter-based parameter-efficient adaptation
4. SageMaker Model Registry

Correct Answer: C

 

Correct Answer

Answer C is correct because the description directly matches LoRA or adapter-based parameter-efficient adaptation. LoRA or adapter-based parameter-efficient adaptation adds or updates compact adaptation parameters while leaving most base-model weights unchanged.

Incorrect Answers

Answer A is incorrect because model cascading is primarily used to use inexpensive or specialized models for routine requests while reserving larger models for harder requests, which is a different requirement from the one being tested.

Answer B is incorrect because foundation model benchmark and capability evaluation is primarily used to select a model by comparing quality, modality, context limits, latency, cost, and task-specific performance, which is a different requirement from the one being tested.

Answer D is incorrect because SageMaker Model Registry is primarily used to version and govern customized model artifacts across their deployment lifecycle, which is a different requirement from the one being tested.

 

Question 16

The SaaS product team needs to identify a capability with this behavior: tracks model versions, metadata, approval state, and promotion decisions for controlled deployment. What is the best match?

1. model abstraction layer with Lambda, API Gateway, and AppConfig
2. automated model deployment with rollback
3. SageMaker Model Registry
4. smaller specialized foundation model

Correct Answer: C

 

Correct Answer

Answer C is correct because the description directly matches SageMaker Model Registry. SageMaker Model Registry tracks model versions, metadata, approval state, and promotion decisions for controlled deployment.

Incorrect Answers

Answer A is incorrect because model abstraction layer with Lambda, API Gateway, and AppConfig is primarily used to switch models or providers without rewriting application business logic, which is a different requirement from the one being tested.

Answer B is incorrect because automated model deployment with rollback is primarily used to update customized models safely and recover quickly from a failed release, which is a different requirement from the one being tested.

Answer D is incorrect because smaller specialized foundation model is primarily used to reduce latency and inference cost for a narrow task that does not need the capabilities of a larger model, which is a different requirement from the one being tested.

 

Question 17

The customer-service platform describes a component that uses versioned artifacts, deployment automation, health checks, and rollback criteria instead of replacing models manually. Which capability is being described?

1. model inference parameter configuration
2. automated model deployment with rollback
3. Amazon Bedrock Cross-Region Inference
4. model cascading

Correct Answer: B

 

Correct Answer

Answer B is correct because the description directly matches automated model deployment with rollback. automated model deployment with rollback uses versioned artifacts, deployment automation, health checks, and rollback criteria instead of replacing models manually.

Incorrect Answers

Answer A is incorrect because model inference parameter configuration is primarily used to tune response determinism, diversity, and generation behavior for a specific use case, which is a different requirement from the one being tested.

Answer C is incorrect because Amazon Bedrock Cross-Region Inference is primarily used to improve resilience or capacity access when a supported model has regional constraints, which is a different requirement from the one being tested.

Answer D is incorrect because model cascading is primarily used to use inexpensive or specialized models for routine requests while reserving larger models for harder requests, which is a different requirement from the one being tested.

 

Question 18

Within the cloud center of excellence’s architecture, which capability matches this technical description: routes work through a sequence or hierarchy of models based on complexity, confidence, or policy?

1. foundation model benchmark and capability evaluation
2. Amazon SageMaker AI endpoint for a customized model
3. model cascading
4. smaller specialized foundation model

Correct Answer: C

 

Correct Answer

Answer C is correct because the description directly matches model cascading. model cascading routes work through a sequence or hierarchy of models based on complexity, confidence, or policy.

Incorrect Answers

Answer A is incorrect because foundation model benchmark and capability evaluation is primarily used to select a model by comparing quality, modality, context limits, latency, cost, and task-specific performance, which is a different requirement from the one being tested.

Answer B is incorrect because Amazon SageMaker AI endpoint for a customized model is primarily used to serve a domain-specific or fine-tuned model that requires managed custom model hosting, which is a different requirement from the one being tested.

Answer D is incorrect because smaller specialized foundation model is primarily used to reduce latency and inference cost for a narrow task that does not need the capabilities of a larger model, which is a different requirement from the one being tested.

 

Question 19

The contact-center transformation team describes a component that selects the smallest model that still satisfies measured quality and functional requirements. Which capability is being described?

1. model inference parameter configuration
2. LoRA or adapter-based parameter-efficient adaptation
3. model abstraction layer with Lambda, API Gateway, and AppConfig
4. smaller specialized foundation model

Correct Answer: D

 

Correct Answer

Answer D is correct because the description directly matches smaller specialized foundation model. smaller specialized foundation model selects the smallest model that still satisfies measured quality and functional requirements.

Incorrect Answers

Answer A is incorrect because model inference parameter configuration is primarily used to tune response determinism, diversity, and generation behavior for a specific use case, which is a different requirement from the one being tested.

Answer B is incorrect because LoRA or adapter-based parameter-efficient adaptation is primarily used to customize model behavior with fewer trainable parameters than full fine-tuning, which is a different requirement from the one being tested.

Answer C is incorrect because model abstraction layer with Lambda, API Gateway, and AppConfig is primarily used to switch models or providers without rewriting application business logic, which is a different requirement from the one being tested.

 

Question 20

The customer analytics platform documents this GenAI behavior: adjusts supported inference parameters such as temperature, top-p, top-k, stop conditions, and maximum output length. Which capability matches it?

1. SageMaker Model Registry
2. model inference parameter configuration
3. foundation model benchmark and capability evaluation
4. Amazon Bedrock Cross-Region Inference

Correct Answer: B

 

Correct Answer

Answer B is correct because the description directly matches model inference parameter configuration. model inference parameter configuration adjusts supported inference parameters such as temperature, top-p, top-k, stop conditions, and maximum output length.

Incorrect Answers

Answer A is incorrect because SageMaker Model Registry is primarily used to version and govern customized model artifacts across their deployment lifecycle, which is a different requirement from the one being tested.

Answer C is incorrect because foundation model benchmark and capability evaluation is primarily used to select a model by comparing quality, modality, context limits, latency, cost, and task-specific performance, which is a different requirement from the one being tested.

Answer D is incorrect because Amazon Bedrock Cross-Region Inference is primarily used to improve resilience or capacity access when a supported model has regional constraints, which is a different requirement from the one being tested.

 

Question 21

The regulatory reporting platform has proposed foundation model benchmark and capability evaluation for its design. Which requirement best justifies it?

1. Update customized models safely and recover quickly from a failed release
2. Switch models or providers without rewriting application business logic
3. Serve a domain-specific or fine-tuned model that requires managed custom model hosting
4. Select a model by comparing quality, modality, context limits, latency, cost, and task-specific performance

Correct Answer: D

 

Correct Answer

Answer D is correct because foundation model benchmark and capability evaluation is specifically used to select a model by comparing quality, modality, context limits, latency, cost, and task-specific performance. It uses representative prompts and measurable acceptance criteria rather than choosing a model only by size or popularity.

Incorrect Answers

Answer A is incorrect because that requirement aligns with automated model deployment with rollback, not foundation model benchmark and capability evaluation.

Answer B is incorrect because that requirement aligns with model abstraction layer with Lambda, API Gateway, and AppConfig, not foundation model benchmark and capability evaluation.

Answer C is incorrect because that requirement aligns with Amazon SageMaker AI endpoint for a customized model, not foundation model benchmark and capability evaluation.

 

Question 22

The privacy engineering function is considering model abstraction layer with Lambda, API Gateway, and AppConfig. What problem is this choice primarily meant to solve?

1. Switch models or providers without rewriting application business logic
2. Improve resilience or capacity access when a supported model has regional constraints
3. Customize model behavior with fewer trainable parameters than full fine-tuning
4. Use inexpensive or specialized models for routine requests while reserving larger models for harder requests

Correct Answer: A

 

Correct Answer

Answer A is correct because model abstraction layer with Lambda, API Gateway, and AppConfig is specifically used to switch models or providers without rewriting application business logic. It places model selection and configuration behind a stable application interface and externalized configuration.

Incorrect Answers

Answer B is incorrect because that requirement aligns with Amazon Bedrock Cross-Region Inference, not model abstraction layer with Lambda, API Gateway, and AppConfig.

Answer C is incorrect because that requirement aligns with LoRA or adapter-based parameter-efficient adaptation, not model abstraction layer with Lambda, API Gateway, and AppConfig.

Answer D is incorrect because that requirement aligns with model cascading, not model abstraction layer with Lambda, API Gateway, and AppConfig.

 

Question 23

Why would the logistics optimization team introduce Amazon Bedrock Cross-Region Inference into the GenAI architecture?

1. Serve a domain-specific or fine-tuned model that requires managed custom model hosting
2. Improve resilience or capacity access when a supported model has regional constraints
3. Version and govern customized model artifacts across their deployment lifecycle
4. Reduce latency and inference cost for a narrow task that does not need the capabilities of a larger model

Correct Answer: B

 

Correct Answer

Answer B is correct because Amazon Bedrock Cross-Region Inference is specifically used to improve resilience or capacity access when a supported model has regional constraints. It routes eligible Bedrock inference requests across supported regions according to a cross-region inference profile.

Incorrect Answers

Answer A is incorrect because that requirement aligns with Amazon SageMaker AI endpoint for a customized model, not Amazon Bedrock Cross-Region Inference.

Answer C is incorrect because that requirement aligns with SageMaker Model Registry, not Amazon Bedrock Cross-Region Inference.

Answer D is incorrect because that requirement aligns with smaller specialized foundation model, not Amazon Bedrock Cross-Region Inference.

 

Question 24

The service reliability organization has proposed Amazon SageMaker AI endpoint for a customized model for its design. Which requirement best justifies it?

1. Tune response determinism, diversity, and generation behavior for a specific use case
2. Serve a domain-specific or fine-tuned model that requires managed custom model hosting
3. Update customized models safely and recover quickly from a failed release
4. Customize model behavior with fewer trainable parameters than full fine-tuning

Correct Answer: B

 

Correct Answer

Answer B is correct because Amazon SageMaker AI endpoint for a customized model is specifically used to serve a domain-specific or fine-tuned model that requires managed custom model hosting. It deploys a controlled model artifact behind a scalable managed inference endpoint.

Incorrect Answers

Answer A is incorrect because that requirement aligns with model inference parameter configuration, not Amazon SageMaker AI endpoint for a customized model.

Answer C is incorrect because that requirement aligns with automated model deployment with rollback, not Amazon SageMaker AI endpoint for a customized model.

Answer D is incorrect because that requirement aligns with LoRA or adapter-based parameter-efficient adaptation, not Amazon SageMaker AI endpoint for a customized model.

 

Question 25

The financial-services engineering group includes LoRA or adapter-based parameter-efficient adaptation in its architecture. What is its primary role?

1. Customize model behavior with fewer trainable parameters than full fine-tuning
2. Version and govern customized model artifacts across their deployment lifecycle
3. Use inexpensive or specialized models for routine requests while reserving larger models for harder requests
4. Select a model by comparing quality, modality, context limits, latency, cost, and task-specific performance

Correct Answer: A

 

Correct Answer

Answer A is correct because LoRA or adapter-based parameter-efficient adaptation is specifically used to customize model behavior with fewer trainable parameters than full fine-tuning. It adds or updates compact adaptation parameters while leaving most base-model weights unchanged.

Incorrect Answers

Answer B is incorrect because that requirement aligns with SageMaker Model Registry, not LoRA or adapter-based parameter-efficient adaptation.

Answer C is incorrect because that requirement aligns with model cascading, not LoRA or adapter-based parameter-efficient adaptation.

Answer D is incorrect because that requirement aligns with foundation model benchmark and capability evaluation, not LoRA or adapter-based parameter-efficient adaptation.

 

Question 26

The software delivery organization is considering SageMaker Model Registry. What problem is this choice primarily meant to solve?

1. Version and govern customized model artifacts across their deployment lifecycle
2. Update customized models safely and recover quickly from a failed release
3. Reduce latency and inference cost for a narrow task that does not need the capabilities of a larger model
4. Switch models or providers without rewriting application business logic

Correct Answer: A

 

Correct Answer

Answer A is correct because SageMaker Model Registry is specifically used to version and govern customized model artifacts across their deployment lifecycle. It tracks model versions, metadata, approval state, and promotion decisions for controlled deployment.

Incorrect Answers

Answer B is incorrect because that requirement aligns with automated model deployment with rollback, not SageMaker Model Registry.

Answer C is incorrect because that requirement aligns with smaller specialized foundation model, not SageMaker Model Registry.

Answer D is incorrect because that requirement aligns with model abstraction layer with Lambda, API Gateway, and AppConfig, not SageMaker Model Registry.

 

Question 27

Within the claims processing service’s design, the team highlights automated model deployment with rollback. Which need does that component address?

1. Tune response determinism, diversity, and generation behavior for a specific use case
2. Improve resilience or capacity access when a supported model has regional constraints
3. Use inexpensive or specialized models for routine requests while reserving larger models for harder requests
4. Update customized models safely and recover quickly from a failed release

Correct Answer: D

 

Correct Answer

Answer D is correct because automated model deployment with rollback is specifically used to update customized models safely and recover quickly from a failed release. It uses versioned artifacts, deployment automation, health checks, and rollback criteria instead of replacing models manually.

Incorrect Answers

Answer A is incorrect because that requirement aligns with model inference parameter configuration, not automated model deployment with rollback.

Answer B is incorrect because that requirement aligns with Amazon Bedrock Cross-Region Inference, not automated model deployment with rollback.

Answer C is incorrect because that requirement aligns with model cascading, not automated model deployment with rollback.

 

Question 28

The clinical knowledge application has proposed model cascading for its design. Which requirement best justifies it?

1. Select a model by comparing quality, modality, context limits, latency, cost, and task-specific performance
2. Reduce latency and inference cost for a narrow task that does not need the capabilities of a larger model
3. Serve a domain-specific or fine-tuned model that requires managed custom model hosting
4. Use inexpensive or specialized models for routine requests while reserving larger models for harder requests

Correct Answer: D

 

Correct Answer

Answer D is correct because model cascading is specifically used to use inexpensive or specialized models for routine requests while reserving larger models for harder requests. It routes work through a sequence or hierarchy of models based on complexity, confidence, or policy.

Incorrect Answers

Answer A is incorrect because that requirement aligns with foundation model benchmark and capability evaluation, not model cascading.

Answer B is incorrect because that requirement aligns with smaller specialized foundation model, not model cascading.

Answer C is incorrect because that requirement aligns with Amazon SageMaker AI endpoint for a customized model, not model cascading.

 

Question 29

Within the mobile experience engineering team’s design, the team highlights smaller specialized foundation model. Which need does that component address?

1. Customize model behavior with fewer trainable parameters than full fine-tuning
2. Reduce latency and inference cost for a narrow task that does not need the capabilities of a larger model
3. Switch models or providers without rewriting application business logic
4. Tune response determinism, diversity, and generation behavior for a specific use case

Correct Answer: B

 

Correct Answer

Answer B is correct because smaller specialized foundation model is specifically used to reduce latency and inference cost for a narrow task that does not need the capabilities of a larger model. It selects the smallest model that still satisfies measured quality and functional requirements.

Incorrect Answers

Answer A is incorrect because that requirement aligns with LoRA or adapter-based parameter-efficient adaptation, not smaller specialized foundation model.

Answer C is incorrect because that requirement aligns with model abstraction layer with Lambda, API Gateway, and AppConfig, not smaller specialized foundation model.

Answer D is incorrect because that requirement aligns with model inference parameter configuration, not smaller specialized foundation model.

 

Question 30

The cloud application engineering team plans to adopt model inference parameter configuration. Which outcome should drive that decision?

1. Select a model by comparing quality, modality, context limits, latency, cost, and task-specific performance
2. Tune response determinism, diversity, and generation behavior for a specific use case
3. Version and govern customized model artifacts across their deployment lifecycle
4. Improve resilience or capacity access when a supported model has regional constraints

Correct Answer: B

 

Correct Answer

Answer B is correct because model inference parameter configuration is specifically used to tune response determinism, diversity, and generation behavior for a specific use case. It adjusts supported inference parameters such as temperature, top-p, top-k, stop conditions, and maximum output length.

Incorrect Answers

Answer A is incorrect because that requirement aligns with foundation model benchmark and capability evaluation, not model inference parameter configuration.

Answer C is incorrect because that requirement aligns with SageMaker Model Registry, not model inference parameter configuration.

Answer D is incorrect because that requirement aligns with Amazon Bedrock Cross-Region Inference, not model inference parameter configuration.