AWS Certified Generative AI Developer - Professional (AIP-C01) (AIP-C01) AIP-C01 Questions 1–75 | Page 1/3

MCQmedium

In SageMaker JumpStart, what is the primary advantage of selecting a 'Provisioned Throughput' option for a deployed model?

A.It automates the model fine-tuning process

B.It reduces the size of the model weights

C.It allows the model to access the internet

D.It ensures dedicated capacity for the endpoint

AnswerD

Provisioning guarantees the requested resources.

Why this answer

Provisioned Throughput guarantees capacity for your model, preventing performance fluctuations during high load.

Full explanation →

MCQmedium

Which component in Amazon Bedrock is used to define the structure of the retrieved documents before they are fed into the model?

A.Prompt Template.

B.Vector Database index settings.

C.Chunking Strategy.

D.IAM Policy.

AnswerC

Chunking defines the size and structure of the input segments.

Why this answer

The 'Chunking Strategy' determines how documents are broken down and structured before being embedded and indexed.

Full explanation →

MCQeasy

Which AWS tool enables you to test different models side-by-side using the same prompt?

A.Amazon SageMaker Studio

B.Bedrock Playground

C.AWS Cloud9

D.AWS Lambda Console

AnswerB

The Playground supports multi-model comparison.

Why this answer

The Bedrock Playground is the integrated environment for testing prompts across multiple models.

Full explanation →

MCQmedium

A financial firm needs to use a third-party foundation model on Amazon Bedrock but must ensure that the model provider does not use their prompts or output to train their base models. Which feature provides this guarantee?

A.AWS PrivateLink

B.AWS KMS Customer Managed Keys

C.Amazon Bedrock Data Privacy Guarantees

D.Amazon Bedrock Provisioned Throughput

AnswerC

AWS explicitly states that inputs and outputs are not used for model training.

Why this answer

AWS Bedrock provides a 'Zero Data Retention' policy by default, ensuring data is not used for model training.

Full explanation →

MCQeasy

When invoking models in Bedrock, what does the 'streaming' feature provide?

A.Support for larger context windows

B.Reduced latency until the first token appears

C.Higher model accuracy

D.Lower cost per invocation

AnswerB

Streaming minimizes perceived wait time.

Why this answer

Streaming allows the application to receive and display tokens as they are generated, improving user experience.

Full explanation →

MCQmedium

Your organization needs to ensure that only authorized IAM users can access specific foundation models in Bedrock. Which configuration is required?

A.Attach an IAM policy with bedrock:InvokeModel allowing access to specific model ARNs

B.Enable AWS Organizations SCPs only

C.Use Amazon Cognito user pools

D.Use Resource-Based Policies on the foundation model

AnswerA

This is the correct approach to restrict access to specific models using IAM.

Why this answer

You must explicitly grant permission to the specific Bedrock model ARNs in the IAM policy attached to the user or role.

Full explanation →

MCQhard

You are using SageMaker JumpStart and need to enable 'Data Capture' for monitoring model inputs and outputs. Where is the captured data stored?

A.Amazon RDS

B.Amazon CloudWatch Logs

C.Amazon ElastiCache

D.Amazon S3

AnswerD

S3 is the designated store for captured inference logs.

Why this answer

SageMaker Data Capture automatically uploads captured request and response payloads to an S3 bucket.

Full explanation →

Multi-Selectmedium

Which TWO factors are critical when choosing a model for a RAG architecture that requires high grounding accuracy?

Select 2 answers

A.The model's training date.

B.The model's ability to generate images.

C.The model's support for citation and source attribution.

D.The number of parameters in the model.

E.The context window size of the foundation model.

AnswersC, E

Citation capability is crucial for verifying the grounding of the model response.

Why this answer

Accuracy in RAG relies on model reasoning capability and the quality of the retrieved chunks.

Full explanation →

MCQeasy

A team is evaluating different models on Amazon Bedrock. They want to compare the performance of Claude 3 Sonnet versus Llama 3 on a specific prompt set without writing custom scripts. Which tool should they use?

A.Amazon SageMaker Model Monitor

B.Amazon QuickSight

C.Amazon Bedrock Prompt Management and Playgrounds

D.AWS Lambda functions

AnswerC

Playgrounds allow for side-by-side comparison of different models and configurations.

Why this answer

Bedrock Prompt Management and Playgrounds provide a GUI to test and compare models directly.

Full explanation →

MCQhard

A firm wants to audit the specific version of a foundation model used for every inference request. How can they achieve this?

A.Query Amazon SageMaker Model Monitor

B.Enable CloudTrail logs

C.Enable Bedrock Model Invocation Logging

D.Use AWS Config to track model changes

AnswerC

This feature logs the inputs, outputs, and metadata including the specific model ID used.

Why this answer

Model Invocation Logging logs the modelId, which includes the version or ARN, providing an audit trail for every request.

Full explanation →

MCQhard

A company is using RAG with Amazon Bedrock Knowledge Bases. They need to ensure that users can only retrieve documents they are authorized to see. How should this be implemented?

A.Encrypt the documents with different KMS keys

B.Create separate Knowledge Bases for every user

C.Apply IAM policies to the S3 bucket

D.Use the filters attribute in the Retrieve API call based on document metadata

AnswerD

The filter functionality in Retrieve allows restricting results based on metadata attributes like 'user_department'.

Why this answer

You must implement metadata filtering in the Knowledge Base query. When invoking Retrieve, you can apply filters based on user attributes stored in the document metadata.

Full explanation →

MCQmedium

You have a fleet of SageMaker endpoints. You want to implement an automatic cost reduction strategy during off-peak hours for specific endpoints. Which approach is most efficient?

A.Using spot instances for all endpoints

B.Configuring Application Auto Scaling with scheduled actions

C.Deactivating the endpoint manually

D.Updating the endpoint configuration

AnswerB

Scheduled actions can change the instance count based on a recurring timeline.

Why this answer

Using a scheduled Auto Scaling policy allows you to reduce the instance count to the minimum required during off-peak hours automatically.

Full explanation →

Multi-Selectmedium

A company is setting up a secure RAG pipeline. Which THREE of the following are necessary to ensure data security?

Select 3 answers

A.Implement IAM policies with the principle of least privilege.

B.Use VPC endpoints for Amazon Bedrock and OpenSearch Serverless.

C.Enable KMS encryption for the vector database and S3 buckets.

D.Use public S3 endpoints with read-only access.

E.Use default AWS managed public keys.

AnswersA, B, C

Essential for security.

Why this answer

VPC endpoints, IAM policies, and KMS encryption are fundamental components for securing data pipelines in AWS.

Full explanation →

Multi-Selecteasy

Which TWO actions should you perform to identify cost savings in an existing SageMaker environment?

Select 2 answers

A.Delete all data in S3

B.Review AWS Trusted Advisor recommendations

C.Analyze spending patterns in AWS Cost Explorer

D.Increase the number of IAM roles

E.Deploy models to every region

AnswersB, C

Trusted Advisor provides specific cost optimization advice.

Why this answer

Analyzing Cost Explorer and using Trusted Advisor are the correct ways to find optimization opportunities.

Full explanation →

Multi-Selectmedium

Which TWO steps are required to properly secure Bedrock Knowledge Bases?

Select 2 answers

A.Disable all encryption to increase performance

B.Place the Knowledge Base in a public subnet

C.Use IAM roles for the Knowledge Base to access data

D.Grant public read access to the S3 bucket

E.Specify a KMS key for encryption at rest

AnswersC, E

Ensures controlled access.

Why this answer

Enabling encryption and managing IAM access are the primary security configurations.

Full explanation →

Multi-Selecthard

Which THREE metrics are best used for setting up auto-scaling policies on a SageMaker endpoint?

Select 3 answers

A.IAMUserLoginCount

B.InvocationsPerInstance

C.CPUUtilization

D.S3RequestLatency

E.GPUUtilization

AnswersB, C, E

A direct measure of traffic load per instance.

Why this answer

InvocationsPerInstance, CPUUtilization, and GPUUtilization are the standard metrics to trigger scaling actions.

Full explanation →

MCQeasy

Which AWS storage service is the primary source for synchronizing documents in an Amazon Bedrock Knowledge Base?

A.Amazon EFS

B.AWS Backup

C.Amazon DynamoDB

D.Amazon S3

AnswerD

S3 is the natively supported source for Knowledge Base synchronization.

Why this answer

Amazon S3 is the standard storage location for documents that are indexed by Bedrock Knowledge Bases.

Full explanation →

MCQhard

A developer is building a RAG system and wants to include document metadata in the search retrieval. How is this done?

A.By prepending the metadata to the prompt.

B.By adding metadata files alongside the documents in S3 and using custom ingestion logic.

C.It is done automatically by the model.

D.It is not possible to use metadata in Bedrock RAG.

AnswerB

This is the standard way to map metadata to chunks.

Why this answer

Metadata is attached to the chunks in the vector index. During query, the system can use the `filter` parameter to limit results.

Full explanation →

MCQhard

You are integrating a SageMaker JumpStart model via the InvokeEndpoint API. The model is failing to process images due to a schema mismatch. Which format does the JumpStart image input expect by default for standard vision models?

A.Base64 encoded string in the JSON input field

B.A URL pointing to an S3 object

C.Multi-part form data

D.Raw binary stream in the request body

AnswerA

Standard JumpStart vision models require base64 encoding.

Why this answer

JumpStart vision models typically expect the image to be base64 encoded within the JSON payload.

Full explanation →

Multi-Selecteasy

Which THREE of the following are valid components of an Amazon Bedrock Agent test plan?

Select 3 answers

A.Validating the reasoning path of the agent.

B.Verifying the S3 bucket encryption.

C.Testing the accuracy of tool invocations (Action Groups).

D.Measuring model training time.

E.Ensuring Guardrails are applied to both input and output.

AnswersA, C, E

The agent must be tested for its ability to follow instructions logically.

Why this answer

Testing reasoning, tool invocation accuracy, and safety constraints are critical components of an agent test plan.

Full explanation →

MCQhard

A researcher wants to ensure the model output does not reference external websites. How can this be enforced?

A.Use a custom system prompt to tell the model not to use URLs

B.Configure a Guardrail with a Denied Topics policy

C.Disable model streaming

D.Use IAM to block internet access for the application

AnswerB

A Denied Topics policy can be configured to block content related to external links or websites.

Why this answer

Bedrock Guardrails can be configured with a Word Policy to deny specific patterns or regex that identify URLs, or a topic policy to block external references.

Full explanation →

MCQmedium

A company is building a RAG application using Amazon Bedrock and needs to ensure that the data ingested into the knowledge base is filtered based on user identity. Which approach should the developer use?

A.Use Amazon Bedrock Knowledge Base metadata filtering during the retrieval phase to apply access controls.

B.Use Lambda authorizers to filter the retrieved results after they are returned by the model.

C.Create separate vector indexes for every user role within the Knowledge Base configuration.

D.Implement fine-grained access control using IAM policies on the S3 bucket containing the source data.

AnswerA

Metadata filtering allows precise control over which documents are retrieved during query time based on user-specific attributes.

Why this answer

Knowledge Base filters allow you to restrict retrieval results based on metadata attributes associated with the document chunks.

Full explanation →

MCQeasy

Which of the following is NOT a benefit of using Amazon Bedrock?

A.Serverless architecture.

B.Access to multiple FM providers via a single API.

C.Data privacy and security controls.

D.Full control over the model's source code.

AnswerD

Bedrock is a managed service; users do not have access to the model source code.

Why this answer

Bedrock is a managed service, so it does not require managing the infrastructure of the underlying foundation models.

Full explanation →

MCQmedium

A team is storing sensitive documents for RAG. They need to ensure that the document chunks in the vector database are encrypted at rest. How?

A.Configure the vector database (e.g., OpenSearch) with KMS encryption

B.Use TLS for all connections

C.Use Amazon S3-managed encryption

D.Encrypt the source S3 bucket only

AnswerA

Vector databases must be configured with encryption at rest using KMS keys.

Why this answer

When creating a Knowledge Base in Bedrock, you must specify a KMS key to encrypt the underlying vector store (e.g., OpenSearch Serverless).

Full explanation →

MCQeasy

What is the purpose of 'Foundation Models' (FMs) in the context of Amazon Bedrock?

A.To perform generative tasks like text completion or analysis.

B.To provide a graphical user interface.

C.To manage VPC networking.

D.To store document metadata.

AnswerA

FMs are the generative engines in the Bedrock ecosystem.

Why this answer

FMs are the core LLMs (like Titan or Claude) that perform the text generation, summarization, or embedding tasks.

Full explanation →

MCQmedium

When should a developer choose to fine-tune a model versus using RAG?

A.When you have very little data.

B.When you need to ensure the model does not hallucinate.

C.When the data changes daily.

D.When you need the model to learn a specific, consistent corporate tone.

AnswerD

Fine-tuning is the correct approach for style/behavior changes.

Why this answer

RAG is best for grounding in changing data, while fine-tuning is better for changing the model's style, tone, or highly specialized domain language.

Full explanation →

MCQmedium

A developer needs to handle documents that are too large to fit in a single prompt. What is the most recommended approach?

A.Compress the text using a ZIP algorithm before sending.

B.Concatenate all text and send it in one prompt.

C.Ask the user to split the document.

D.Use a Knowledge Base to perform retrieval and generation.

AnswerD

This automatically manages chunking and context window limits.

Why this answer

Using a Knowledge Base with a retriever-generator pattern ensures that only relevant chunks are sent to the LLM, effectively managing context windows.

Full explanation →

MCQhard

A developer needs to integrate a foundation model that requires custom fine-tuning for a specific domain language. Which service provides the environment for this task?

A.Amazon Bedrock Custom Models

B.AWS Glue DataBrew

C.Amazon SageMaker JumpStart

D.Amazon Kendra

AnswerA

Custom Models in Bedrock support fine-tuning of supported base models.

Why this answer

Amazon Bedrock Custom Models allow for fine-tuning base models on private datasets to improve performance on specific tasks or domains.

Full explanation →

MCQmedium

A company requires their AI to be available even if a regional outage occurs. How do they handle Bedrock model availability?

A.Implement application-level logic to switch regions for inference.

B.Use an S3 bucket in every region.

C.Request a cross-region Bedrock endpoint from AWS Support.

D.Enable global AWS region auto-scaling.

AnswerA

Applications must handle regional failover for service availability.

Why this answer

Cross-region inference is not supported by all models, so the application logic must handle failover between regions where the model is available.

Full explanation →

MCQmedium

You need to ensure that your SageMaker inference endpoint does not exceed a specific monthly budget. Which AWS feature allows you to set up notifications when spending thresholds are crossed?

A.SageMaker Cost Explorer

B.CloudWatch Alarms on billing metrics

C.AWS Budgets

D.AWS Cost and Usage Reports

AnswerC

AWS Budgets allows proactive threshold monitoring and alerting.

Why this answer

AWS Budgets allows you to set custom cost and usage budgets and receive email or SNS alerts when thresholds are approached or exceeded.

Full explanation →

Multi-Selectmedium

A developer is testing an application and finds that the model output is too verbose. Which TWO of the following parameters or techniques can help control the response length?

Select 2 answers

A.Reduce the 'max_tokens' parameter.

B.Increase the temperature.

C.Enable Guardrails to truncate output.

D.Use a different model family.

E.Update the system prompt to include 'be concise'.

AnswersA, E

This sets a hard limit on the number of generated tokens.

Why this answer

Max tokens and system instructions are the primary ways to limit verbosity.

Full explanation →

MCQmedium

Which metric should a developer monitor to ensure the RAG system remains performant as usage increases?

A.Vector Database Storage Capacity.

B.IAM User Count.

C.Inference Latency.

D.Model Training Loss.

AnswerC

Latency is critical for user experience in RAG applications.

Why this answer

Latency and Token Consumption (or request count) are key indicators of RAG performance and cost efficiency.

Full explanation →

MCQmedium

An organization wants to use a model that supports both text and image input. Which family of foundation models should they select in Amazon Bedrock?

A.Amazon Titan Text Express

B.Claude 3 Sonnet

C.Jurassic-2 Mid

D.Cohere Command R

AnswerB

Claude 3 models are multi-modal and accept image inputs.

Why this answer

Amazon Bedrock supports multi-modal models like Claude 3 Sonnet and Claude 3 Opus which handle both text and images.

Full explanation →

MCQeasy

What is the primary role of an Embedding Model in a RAG pipeline?

A.To convert text into numerical vectors for semantic similarity search.

B.To act as a firewall for the RAG system.

C.To compress large files for storage.

D.To generate human-readable summaries.

AnswerA

This is the core function of an embedding model.

Why this answer

Embedding models convert text into numerical vectors that represent the semantic meaning of the content for vector search.

Full explanation →

MCQmedium

You are building a RAG application. Which metadata field is most useful for filtering results during retrieval?

A.Vector distance

B.Token count

C.Model ID

D.Creation date

AnswerD

Very common for filtering, though category is also used.

Why this answer

Metadata filtering allows you to restrict retrieval to specific document types or categories.

Full explanation →

Multi-Selectmedium

Which TWO types of models are generally available in Amazon Bedrock?

Select 2 answers

A.Embedding models.

B.Network load balancer configurations.

C.Hardware drivers.

D.OS kernel images.

E.Text generation models.

AnswersA, E

Core functionality.

Why this answer

Text generation models and embedding models are the two most common types available in Bedrock.

Full explanation →

MCQmedium

When designing a RAG system, why would a developer choose Amazon Titan Text Embeddings over a third-party embedding model?

A.It is optimized for Amazon Bedrock Knowledge Bases, ensuring seamless integration and scalability.

B.It supports multi-modal input processing natively.

C.It is the only model that supports English.

D.It prevents the model from hallucinating.

AnswerA

Natively integrated models reduce architectural overhead and latency.

Why this answer

Amazon Titan Embeddings are optimized to work natively with Amazon Bedrock Knowledge Bases and provide high performance and integration benefits.

Full explanation →

MCQmedium

A company is integrating Amazon Bedrock with a private knowledge base. They need to ensure that PII is masked before the retrieved documents are passed to the Foundation Model. Which approach is most effective?

A.Configure the Bedrock Knowledge Base to use a custom encryption key for PII filtering.

B.Enable Amazon Macie on the S3 bucket and configure Bedrock to ignore flagged objects.

C.Implement an AWS Lambda function with Amazon Comprehend to redact PII during the ingestion process into the Knowledge Base.

D.Use Amazon Bedrock Guardrails to mask PII during the inference request.

AnswerC

Pre-processing data using AWS Lambda and Amazon Comprehend before ingestion ensures sensitive data never reaches the vector index.

Why this answer

Amazon Comprehend can be integrated into the data ingestion pipeline to detect and redact PII before documents are indexed into the Amazon OpenSearch Serverless vector store.

Full explanation →

MCQhard

You notice your SageMaker endpoint is underperforming. You identify that the model is CPU-bound during inference. Which action should you take?

A.Reduce the batch size

B.Increase memory limit in the container

C.Increase the number of instances (horizontal scaling)

D.Change to an instance type optimized for compute

AnswerD

Vertical scaling targets the bottleneck resource.

Why this answer

Upgrading to an instance type with more compute (CPU) resources is the correct path for CPU-bound workloads.

Full explanation →

Multi-Selecthard

When fine-tuning a model on Amazon Bedrock, which THREE security considerations are critical?

Select 3 answers

A.Using a root user account to launch the training job

B.Limiting IAM permissions for the fine-tuning job

C.Encrypting the training dataset in S3 with KMS

D.Making the training dataset publicly readable

E.Ensuring the training dataset is free of PII

AnswersB, C, E

Ensures only authorized users can trigger jobs.

Why this answer

Data governance, encryption, and access control are critical for fine-tuning security.

Full explanation →

MCQeasy

A company is building a RAG application on Amazon Bedrock. They need to store embeddings for a large-scale knowledge base with low-latency similarity search and full integration with LangChain. Which vector database should the developer recommend?

A.Amazon DynamoDB with global secondary indexes

B.Amazon Aurora PostgreSQL without pgvector

C.Amazon S3 with Athena query engine

D.Amazon OpenSearch Serverless vector engine

AnswerD

OpenSearch Serverless vector engine is purpose-built for high-performance vector search in GenAI applications.

Why this answer

Amazon OpenSearch Serverless with the vector engine is the native choice for Bedrock RAG workflows, offering seamless integration and scalability.

Full explanation →

MCQmedium

Your team uses SageMaker Pipelines for training. How can you optimize the storage costs of the artifacts generated by these runs?

A.Disabling versioning on the S3 bucket

B.Configuring S3 Lifecycle policies

C.Using smaller instance types

D.Deleting logs every day

AnswerB

S3 Lifecycle policies manage artifact retention and tiering automatically.

Why this answer

Implementing an S3 Lifecycle policy automatically moves or deletes artifacts after a certain period, keeping storage costs controlled.

Full explanation →

MCQeasy

Which data format is best suited for ingestion into an Amazon Bedrock Knowledge Base?

A.Raw binary executable files.

B.PDF, Markdown, or HTML files stored in Amazon S3.

C.Encrypted raw stream data.

D.Proprietary database dump files.

AnswerB

These are the supported text-rich formats for ingestion.

Why this answer

Bedrock Knowledge Bases support various formats, but PDF, HTML, and Markdown are standard and natively handled.

Full explanation →

MCQmedium

A developer wants to ensure that a generative AI application does not provide medical advice. How should they configure this?

A.Disable the model's knowledge of medical terms.

B.Instruct the model via the system prompt only.

C.Create a Guardrail with a 'Denied Topics' policy for medical advice.

D.Use a prompt chain to filter out medical terms.

AnswerC

Guardrails provide an enforceable boundary for specific forbidden topics.

Why this answer

Bedrock Guardrails can be used to set a 'Denied Topics' policy, specifically blocking content related to medical advice.

Full explanation →

Multi-Selectmedium

Which TWO metrics are critical for monitoring an Amazon SageMaker inference endpoint to ensure operational health?

Select 2 answers

A.Invocations

B.NetworkPacketLoss

C.S3StorageUsage

D.ModelLatency

E.IAMUserCount

AnswersA, D

Tracks the number of requests to the endpoint.

Why this answer

Invocations and ModelLatency are the most fundamental metrics to gauge load and performance.

Full explanation →

MCQmedium

Which prompt engineering technique involves providing the model with a few examples of desired input-output pairs to guide the generation?

A.Chain-of-thought

B.Few-shot prompting

C.System priming

D.Zero-shot prompting

AnswerB

Few-shot provides examples.

Why this answer

Few-shot prompting provides examples to help the model learn the expected format and style.

Full explanation →

MCQmedium

When using a Knowledge Base, what is the 'Base Prompt' used for?

A.To instruct the model on how to format the answer using retrieved context.

B.To define the database schema.

C.To authenticate the user.

D.To store the conversation history.

AnswerA

This is the instruction set for combining RAG data with the generation task.

Why this answer

The base prompt serves as the template for the LLM to understand how to use retrieved context to answer the user's question.

Full explanation →

MCQeasy

Which Bedrock API allows for the asynchronous ingestion of data into a Knowledge Base?

A.InvokeModel

B.ListKnowledgeBases

C.Retrieve

D.StartIngestionJob

AnswerD

This starts the data synchronization process.

Why this answer

StartIngestionJob is the API operation used to trigger a data sync to a Knowledge Base.

Full explanation →

MCQmedium

A developer needs to ensure that an application only uses approved foundation models. How can they govern this centrally?

A.Use Service Control Policies (SCPs) to deny non-approved models

B.Create an IAM role for each model

C.Use Amazon Cognito

D.Hardcode the model ID in the application code

AnswerA

SCPs allow for fine-grained control over which API actions (and specific model ARNs) are permitted.

Why this answer

Using AWS Organizations SCPs, an administrator can restrict the bedrock:InvokeModel action to a list of allowed model IDs.

Full explanation →

MCQmedium

An organization is concerned about data residency requirements for their AI models. Where should they look to identify which AWS regions support specific models?

A.The IAM console.

B.The AWS CLI.

C.The Amazon S3 bucket settings.

D.The Amazon Bedrock documentation region table.

AnswerD

This document specifies model availability per region.

Why this answer

The AWS Region Table for Amazon Bedrock lists the models available in each region.

Full explanation →

MCQmedium

What is the primary benefit of deploying a JumpStart model on a Multi-Model Endpoint (MME)?

A.It improves inference latency for a single model

B.It simplifies the model training process

C.It optimizes resource utilization by hosting multiple models

D.It provides higher availability than single-model endpoints

AnswerC

Cost-efficiency is the key benefit.

Why this answer

MMEs allow you to host multiple models on a single endpoint, sharing hardware resources and reducing cost.

Full explanation →

MCQmedium

A company is building a RAG application using Amazon Bedrock and needs to ensure that sensitive PII is filtered out before it reaches the foundation model. Which feature should be used to achieve this?

A.AWS WAF on the Bedrock API endpoint

B.Amazon Macie on the S3 bucket housing the source data

C.Amazon Bedrock Guardrails

D.AWS Glue DataBrew

AnswerC

Guardrails provides PII redaction and content filtering capabilities.

Why this answer

Amazon Bedrock Guardrails allows users to configure content filters, including PII redaction, to prevent sensitive information from being processed by or returned from foundation models.

Full explanation →

Multi-Selectmedium

Which THREE features are provided by Amazon Bedrock Guardrails?

Select 3 answers

A.Automatic Model Fine-tuning

B.Real-time cost optimization

C.Content Filtering

D.Denied Topic Management

E.Sensitive Information Filtering

AnswersC, D, E

Filters harmful content.

Why this answer

Guardrails provide sensitive information filtering, content filtering, and denied topic management.

Full explanation →

MCQhard

A business requirement is to ensure that the RAG model output is grounded solely in the provided documents. How can this be enforced?

A.Increase the temperature of the model to 1.0.

B.Use a secondary model to re-check the output.

C.Set the model temperature to a very low value (e.g., 0.0) and use clear system instructions.

D.Disable the model's training data.

AnswerC

Low temperature minimizes randomness and forces the model to stick to the provided context.

Why this answer

By configuring the prompt template with strict instructions and utilizing the 'RetrieveAndGenerate' API with the 'source-only' constraint (where supported), you force grounding.

Full explanation →

MCQeasy

Which service should be used to monitor the cost and usage of API requests made to Amazon Bedrock models?

A.Amazon CloudWatch Metrics

B.AWS Cost Explorer

C.AWS Trusted Advisor

D.Amazon QuickSight

AnswerB

Cost Explorer allows for fine-grained analysis of AWS spending.

Why this answer

AWS Cost Explorer provides detailed analytics and reporting on usage costs for all AWS services, including Bedrock.

Full explanation →

MCQmedium

A developer wants to ensure that the RAG pipeline is resilient to outages. What architectural pattern should be used?

A.Single-node deployment.

B.Hardcoding API endpoints.

C.Using only one S3 bucket.

D.Multi-region deployment with cross-region replication for the vector database.

AnswerD

Cross-region resilience is required for high-availability production workloads.

Why this answer

Deploying the RAG application in multiple availability zones and using regional replication for the vector store is the standard resiliency pattern.

Full explanation →

MCQmedium

Which parameter in Bedrock inference controls the diversity of the generated tokens?

A.Max_tokens

B.Temperature

C.Top_k

D.Top_p

AnswerD

Top_p is the standard for diversity control.

Why this answer

Top_p (also called nucleus sampling) limits the token choices to the top percentage of probability mass, affecting diversity.

Full explanation →

MCQmedium

A company is concerned about model provider policy compliance. Where can they find the specific data processing terms for their chosen model?

A.In the Amazon Bedrock console 'Model Evaluation' tab.

B.By asking the model during an inference session.

C.In the Amazon CloudWatch monitoring dashboard.

D.In the AWS Service Terms and individual model provider documentation.

AnswerD

Legal and privacy terms are detailed in the official service agreements.

Why this answer

The AWS Service Terms and the Bedrock documentation page specify the data handling and privacy commitments for each model provider.

Full explanation →

MCQhard

An enterprise requires that all Generative AI model outputs are logged for compliance, but logs must not contain any PII. How can this be achieved in an AWS environment using Amazon Bedrock?

A.Write a Lambda function to strip logs before writing to S3

B.Enable Bedrock Model Invocation Logging with Guardrails PII redaction enabled

C.Use IAM policies to prevent logs from being written

D.Use AWS CloudTrail to log API calls and enable redaction in CloudWatch

AnswerB

This is the native method to log inference data while ensuring PII is stripped by the Guardrail.

Why this answer

Configuring Model Invocation Logging in Bedrock to send logs to CloudWatch or S3, combined with Guardrails PII redaction, ensures compliant logging.

Full explanation →

Multi-Selectmedium

You are designing a secure Generative AI architecture on AWS. Which TWO of the following are recommended practices for securing the data ingestion pipeline for RAG?

Select 2 answers

A.Disable all logging in the pipeline to prevent data exposure

B.Store API keys in plain text within the Lambda code

C.Enforce TLS 1.2+ for all data movement

D.Use AWS KMS to encrypt data at rest in the S3 bucket

E.Store all documents in a public S3 bucket for ease of access

AnswersC, D

This secures data in transit.

Why this answer

Encrypting data in transit (TLS) and at rest (KMS) are fundamental for data pipeline security.

Full explanation →

MCQeasy

To prevent accidental deletion of a Knowledge Base, which feature should be applied?

A.Use AWS Backup

B.Enable MFA Delete on S3

C.Enable S3 Versioning

D.Apply an IAM policy denying DeleteKnowledgeBase

AnswerD

Restricting the Delete action via IAM is the standard way to protect AWS resources.

Why this answer

AWS resource policies or IAM policies can be used to deny the 'Delete' action on the Knowledge Base resource.

Full explanation →

MCQmedium

A developer is using Amazon Bedrock to build a customer support chatbot. The model is hallucinating frequently when retrieving internal company knowledge. Which evaluation methodology should the developer implement to quantify the ground truth accuracy of the RAG pipeline?

A.Implement a RAGAS-based evaluation pipeline to measure faithfulness and answer relevance.

B.Increase the temperature setting to 1.0 to encourage more creative responses.

C.Use Amazon CloudWatch logs to calculate the latency of every inference call.

D.Run a standard BLEU score evaluation against the model's training data.

AnswerA

RAGAS metrics are industry standard for measuring the effectiveness of the retrieval and generation components.

Why this answer

Retrieval-Augmented Generation (RAG) evaluation requires metrics like RAGAS or TruLens to measure faithfulness, answer relevance, and context precision against a golden dataset.

Full explanation →

Multi-Selecthard

A healthcare organization must ensure that PII is removed from documents before they are stored in an Amazon Bedrock Knowledge Base. Which THREE actions should the developer take?

Select 3 answers

A.Manually encrypt all files using AWS KMS before uploading to S3.

B.Configure the Knowledge Base to use a private VPC endpoint to prevent data exfiltration.

C.Enable Amazon Bedrock Data Protection policies to automatically strip PII during vectorization.

D.Use Amazon Comprehend to detect and redact PII in the source documents before ingestion.

E.Implement an AWS Glue ETL job to sanitize data during the ingestion pipeline.

AnswersB, D, E

VPC endpoints ensure data traverses the AWS network, which is a compliance requirement.

Why this answer

Data preparation involves identifying, masking, and auditing data before it reaches the vector store.

Full explanation →

Multi-Selecthard

A security auditor requests a report on Bedrock usage. Which TWO of the following AWS services can be used to generate this report?

Select 2 answers

A.AWS CodePipeline.

B.AWS AppConfig.

C.Amazon CloudWatch.

D.AWS CloudTrail.

E.AWS Trusted Advisor.

AnswersC, D

Provides logs and metrics for analysis.

Why this answer

CloudTrail and CloudWatch are the primary services for auditing API usage and logging application events.

Full explanation →

MCQeasy

You are deploying a foundation model on Amazon SageMaker. To optimize inference costs for an application with unpredictable traffic patterns and significant idle periods, which deployment option should you select?

A.SageMaker Serverless Inference

B.Batch Transform

C.Real-time inference with provisioned Auto Scaling instances

D.Asynchronous Inference

AnswerA

Serverless scales automatically and charges only for the duration of request processing.

Why this answer

Serverless Inference is the most cost-effective solution for unpredictable traffic as it eliminates the cost of idle instances by automatically scaling resources based on request volume.

Full explanation →

MCQhard

You are tuning a Large Language Model on SageMaker and notice that GPU memory utilization is consistently low despite high latency. Which action is most appropriate to optimize performance?

A.Increasing the number of instances

B.Switching to a larger instance type

C.Moving the model to CPU instances

D.Implementing dynamic batching in your inference handler

AnswerD

Dynamic batching groups requests to maximize GPU utilization and reduce latency.

Why this answer

Increasing the batch size or throughput via model parallelism or optimizing the inference script allows better utilization of the GPU hardware.

Full explanation →

Multi-Selectmedium

You are designing an Agentic AI solution using Bedrock Agents. Which THREE components must you configure to enable the agent to interact with an external API?

Select 3 answers

A.Knowledge Base

B.Action Group

C.OpenAPI Schema

D.Lambda Function

E.Guardrail

AnswersB, C, D

Required to group actions.

Why this answer

Agents require an action group, an OpenAPI schema, and an associated Lambda function to execute the logic.

Full explanation →

Multi-Selectmedium

When designing a prompt for a coding assistant, which THREE pieces of information should be included to get the best results?

Select 3 answers

A.Frameworks or libraries to be used

B.The programming language

C.The AWS account ID

D.Specific formatting requirements

E.The model's internal parameter count

AnswersA, B, D

Specific libraries influence code style.

Why this answer

Code assistants benefit from clear language, context/libraries, and output constraints.

Full explanation →

Multi-Selectmedium

Which THREE actions can a developer take to mitigate hallucinations in a RAG application?

Select 3 answers

A.Use a clear system prompt that restricts the model to only use provided information.

B.Use the retrieve-only approach and perform generation with a custom internal engine.

C.Set model temperature to 0.

D.Ensure high-quality source documents are used in the knowledge base.

E.Allow the model to use its internal training data for any gap in knowledge.

AnswersA, C, D

Forces grounding.

Why this answer

Lowering temperature, providing context, and defining strict system prompts all help limit the model's creative deviation.

Full explanation →

MCQhard

You are deploying a model from SageMaker JumpStart. You need to ensure that the model inference container has access to private VPC resources. What must you configure?

A.Configure Subnets and Security Groups in the Model creation

B.Set the IAM role to 'AdministratorAccess'

C.Update the JumpStart model Dockerfile

D.Enable AWS PrivateLink on the S3 bucket

AnswerA

VPC configuration is required for container network access.

Why this answer

You must configure the VPC settings (Subnets and Security Groups) in the SageMaker Model configuration to allow network access.

Full explanation →

Multi-Selecteasy

Which TWO of the following strategies are best for reducing the cost of SageMaker training jobs?

Select 2 answers

A.Keeping training clusters running

B.Using Managed Spot Training

C.Increasing the number of instances

D.Selecting the most appropriate instance type for the workload

E.Enabling detailed logging for every epoch

AnswersB, D

Spot instances offer significant cost savings for fault-tolerant training.

Why this answer

Using Managed Spot Training and instance-appropriate types are standard methods for reducing training costs.

Full explanation →

MCQmedium

You are deploying a Generative AI application using Amazon SageMaker JumpStart. You need to ensure that the model endpoint is protected against prompt injection attacks. Which approach is most effective?

A.Encrypt the model weights with a CMK

B.Enable Amazon Inspector on the endpoint

C.Use a prompt-based filter system that detects malicious intent via a secondary validation model

D.Disable the API Gateway execute-api method

AnswerC

Using a secondary model or Guardrails to evaluate input intent is the standard defense against prompt injection.

Why this answer

Implementing Guardrails for Amazon Bedrock or a similar input validation layer using semantic checks is the recommended way to mitigate prompt injection.

Full explanation →

Multi-Selectmedium

A developer needs to monitor their Bedrock application for potential abuse. Which TWO features should they utilize?

Select 2 answers

A.Amazon CloudWatch metrics (e.g., 5xx error rates)

B.AWS Cost Explorer to track daily spending

C.Amazon Simple Notification Service

D.AWS Trusted Advisor performance checks

E.AWS CloudTrail to monitor for unusual API calling patterns

AnswersA, E

High error rates often indicate abuse attempts.

Why this answer

CloudWatch metrics for error rates and CloudTrail for audit logs are the primary methods for detecting abuse.

Full explanation →

Multi-Selecthard

Which THREE services/features are essential for establishing a secure perimeter for an Amazon Bedrock deployment?

Select 3 answers

A.Interface VPC Endpoints (PrivateLink)

B.AWS Public Access Block on all buckets

C.AWS Shield Standard

D.Security Groups on the VPC

E.IAM Policies for resource access control

AnswersA, D, E

Enables private API access.

Why this answer

VPC Endpoints (private connectivity), IAM (access), and Security Groups (network access) are essential.

Full explanation →

MCQeasy

When using the Amazon Bedrock Converse API, which parameter enables the model to perform tool use by providing a definition of available functions?

A.inferenceConfig

B.toolConfig

C.system

D.additionalModelRequestFields

AnswerB

toolConfig is the correct parameter to define tools.

Why this answer

The tools parameter in the Converse API allows you to define a list of tools for the model to use.

Full explanation →

AWS Certified Generative AI Developer - Professional (AIP-C01) (AIP-C01) (AIP-C01) — Questions 1–75