AIF-C01 Practice Test 15 — 15 Questions

Question 1

A machine learning team is building a credit risk model and discovers that the training data has a significant imbalance in loan approval rates between two demographic groups. They decide to reweight the training samples using a preprocessing technique. Which SageMaker Clarify feature can help compute the appropriate sample weights to achieve demographic parity?

Choose an answer to begin — your selection is scored in the full session.

Accepted Answer

Clarify preprocessing (reweighting). SageMaker Clarify's preprocessing (reweighting) feature directly computes sample weights to adjust for imbalances in training data, enabling demographic parity by assigning higher weights to underrepresented groups. This is a pre-training bias mitigation technique that modifies the dataset before model training, aligning with the team's goal of reweighting samples to address loan approval rate disparities.

Answer

Clarify post-training bias metrics

Answer

Model Monitor bias drift

Answer

Clarify explainability (SHAP)

Question 2

A company is deploying a Bedrock application that handles sensitive customer data. They must ensure that personally identifiable information (PII) is not included in model responses. Which THREE features of Bedrock Guardrails can help with this requirement? (Choose THREE)

Accepted Answer

PII detection and redaction to mask or remove PII. Option A is correct because Bedrock Guardrails provides sensitive information filters that detect PII entities (such as names, email addresses, phone numbers, and SSNs) and can either mask them with placeholders or block the response entirely, directly preventing PII from appearing in model outputs. Option E is correct because topic denial (denied topics) lets you define custom topics with names, definitions, and example phrases so Guardrails blocks any response that discusses those subjects, which can be used to prevent the model from venturing into areas where PII might be exposed. Option B is not correct because content filters in Guardrails evaluate prompts and responses against harmful-content categories (hate, insults, sexual, violence, misconduct, and prompt-attack) and do not detect or redact PII. Option C is not correct because vector store encryption is a data-at-rest protection for knowledge bases and does not inspect or filter model responses for PII. Option D is not correct because embedding model selection affects how documents are vectorized for retrieval and has no capability to detect or redact PII in generated responses.

Answer

Content filtering to block harmful or offensive language

Answer

Vector store encryption

Answer

Embedding model selection

Question 3

A company uses Amazon Bedrock to build a generative AI application. They need to secure the application by restricting access to the model and preventing sensitive data from being stored in prompts. Which TWO actions should they take? (Choose two.)

Accepted Answer

Use Amazon Bedrock Guardrails to define content filters and deny topics.. Amazon Bedrock Guardrails (Option C) directly addresses the need to prevent sensitive data from being stored in prompts by allowing you to define content filters, deny topics, and configure sensitive information filters that block or mask such data before it reaches the model. This is the native service feature designed for content safety and data leakage prevention.

Answer

Enable VPC Flow Logs for Bedrock.

Answer

Enable AWS CloudTrail to log all Bedrock API calls.

Answer

Use AWS KMS to encrypt the model data.

Question 4

A company wants to analyze customer feedback to identify common themes and sentiment. The feedback is in multiple languages. Which AWS services should be used together?

Accepted Answer

Amazon Translate and Amazon Comprehend. Amazon Translate provides real-time and batch language translation, enabling the conversion of multilingual customer feedback into a single target language. Amazon Comprehend then performs natural language processing (NLP) on the translated text to identify key themes, entities, and sentiment. Together, they form a complete pipeline for analyzing feedback across multiple languages.

Answer

Amazon Transcribe and Amazon Comprehend

Answer

Amazon Polly and Amazon Translate

Answer

Amazon Rekognition and Amazon Comprehend

Question 5

Which THREE considerations are essential when deploying a generative AI application in a regulated industry such as healthcare?

Accepted Answer

Full audit trail of model inputs and outputs for accountability.. In a regulated industry such as healthcare, option B (full audit trail of model inputs and outputs) is essential because accountability and traceability are required to investigate incidents, demonstrate compliance during audits, and reconstruct how a clinical or patient-facing decision was produced. Option C (robust content filtering) is essential because generative models can hallucinate or emit harmful advice, and filtering/guardrails are needed to prevent unsafe or inaccurate outputs from reaching patients or clinicians. Option E (data privacy and HIPAA compliance) is essential because healthcare data is protected health information, so the deployment must enforce safeguards such as access controls, encryption, and Business Associate Agreements to avoid regulatory violations. Option A is not essential in this context because lowest possible inference latency is a performance optimization, not a regulatory or safety requirement, and accuracy, privacy, and auditability take precedence. Option D is not appropriate because maximum creative freedom increases variability and hallucination risk, which directly conflicts with the determinism, safety, and compliance needs of regulated healthcare use.

Answer

Lowest possible inference latency for real-time responses.

Answer

Maximum creative freedom for the model to generate diverse responses.

Question 6

A company wants to automatically detect anomalies in server metrics. Which algorithm is most appropriate?

Accepted Answer

One-class SVM. One-class SVM is specifically designed for anomaly detection, as it learns a boundary around the normal data points in the feature space and identifies any point falling outside this boundary as an anomaly. This makes it ideal for detecting unusual patterns in server metrics without requiring labeled anomaly examples.

Answer

XGBoost

Answer

Linear SVM

Answer

K-Means

Question 7

Refer to the exhibit. A company sets up a knowledge base for a customer support chatbot using Amazon Bedrock. Users report that the chatbot misses relevant details from long documents. Which change to the data source configuration would most likely improve retrieval?

Accepted Answer

Change chunking strategy to SEMANTIC. Semantic chunking groups text based on meaning rather than fixed token counts, preserving the natural boundaries of concepts and paragraphs. This ensures that relevant details from long documents remain intact within a single chunk, improving retrieval accuracy for the chatbot.

Answer

Increase the chunk size in FIXED_SIZE chunking

Answer

Add more documents to the S3 bucket

Answer

Change the embedding model to a larger one

Question 8

A company needs to ensure that model inference endpoints in SageMaker are only accessible from a private subnet in their VPC, and no traffic goes over the public internet. Which network configuration should they use?

Accepted Answer

Use a VPC interface endpoint for SageMaker and disable public access. To keep SageMaker inference traffic entirely off the public internet, you create a VPC interface endpoint (powered by AWS PrivateLink) for the SageMaker API and runtime, then disable the public endpoint so the service is reachable only from within the VPC. This forces all inference calls to traverse the private endpoint ENI in the subnet, never leaving the AWS network.

Answer

Create a public endpoint and restrict access using IAM policies

Answer

Use AWS PrivateLink for the endpoint, but keep public endpoint enabled

Answer

Deploy the endpoint in a public subnet with a security group blocking all inbound traffic

Question 9

A company is using Amazon Bedrock to build a multilingual support chatbot. They need a model that can understand and generate text in multiple languages without requiring separate fine-tuning per language. Which model capability is MOST important?

Accepted Answer

Multilingual training data. A model's ability to understand and generate text in multiple languages without separate fine-tuning depends on being trained on a diverse, multilingual dataset. This allows the model to learn cross-lingual representations and perform zero-shot transfer between languages, which is essential for a multilingual chatbot.

Answer

Large context window

Answer

Support for streaming responses

Answer

High temperature setting

Question 10

Which TWO of the following are best practices for preparing training data for a machine learning model?

Accepted Answer

Handle missing values by imputing or removing them.. Option A is correct because missing values can bias or break many ML algorithms, so best practice is to address them explicitly—either by imputation (e.g., mean/median/mode or model-based) or by removing affected rows/columns when appropriate. Option B is correct because splitting data into training, validation, and test sets enables unbiased model fitting, hyperparameter tuning, and final performance estimation, preventing data leakage and overfitting. Option C is not a best practice because outliers may be legitimate signal; blindly removing them can distort the distribution and reduce robustness, so they should be investigated and handled contextually. Option D is not a best practice because training on the entire dataset leaves no held-out data for validation or testing, making it impossible to reliably estimate generalization. Option E is not a best practice because shuffling is typically needed to remove ordering bias and ensure representative mini-batches, especially when data is sorted by class or time.

Answer

Remove all outliers to improve model robustness.

Answer

Use the entire dataset for training to maximize data usage.

Answer

Avoid shuffling the data to preserve original order.

Question 11

A developer wants to test different prompt variations for a chatbot without making repeated API calls. Which Amazon Bedrock feature can help compare model responses?

Accepted Answer

Amazon Bedrock Playground. Amazon Bedrock Playground provides an interactive console interface where developers can test and compare different prompt variations, model configurations, and foundation models side-by-side without making repeated API calls. It allows real-time experimentation with parameters like temperature, top-p, and prompt engineering to observe how the model responds, making it the correct choice for this use case.

Answer

Model evaluation on Amazon SageMaker

Answer

AWS Security Token Service (STS)

Answer

Amazon CloudWatch Logs

Question 12

Which of the following is a key principle of responsible AI according to AWS?

Accepted Answer

Transparency. Transparency is a core principle of responsible AI according to AWS, meaning that customers should be able to understand how an AI system makes decisions, including its inputs, outputs, and limitations. AWS emphasizes this through features like model cards in Amazon SageMaker, which document model performance, intended uses, and biases. This principle ensures that AI systems are explainable and auditable, building trust with users.

Answer

Complexity

Answer

Speed

Answer

Profitability

Question 13

A company uses Amazon SageMaker Ground Truth to label a dataset for a binary classifier. To reduce labeling bias, which workforce configuration is most appropriate?

Accepted Answer

Private workforce of domain experts. A private workforce of domain experts ensures that labeling is performed by individuals with deep knowledge of the data domain, which directly reduces labeling bias. Domain experts are less likely to misinterpret ambiguous data points and can apply consistent, informed judgment, thereby minimizing systematic errors that could skew the binary classifier's training data.

Answer

Automatic labeling with Active Learning

Answer

Public workforce with no qualification

Answer

Vendor managed workforce

Question 14

A company is using Bedrock Agents to automate multi-step workflows that interact with external APIs and databases. They need to ensure the agent can perform actions like querying a database and calling an API. Which TWO components must be defined to enable these capabilities? (Choose TWO)

Accepted Answer

Lambda functions that implement the business logic for each action. In Amazon Bedrock Agents, action groups (option C) are the required construct that defines the set of actions the agent can invoke, including the OpenAPI schema describing the external APIs and the parameters/operations for database interactions, so the agent knows what actions exist and how to call them. Lambda functions (option B) provide the actual business logic execution behind those action groups, receiving the agent's invocation and performing the concrete work such as running SQL queries against a database or calling an external API. Together, the action group defines the interface and the Lambda function implements it, which is exactly what is needed for querying a database and calling an API. Option A is incorrect because prompt flows are for orchestrating prompt sequences in Amazon Bedrock Flows, not for defining agent action capabilities. Option D is incorrect because Guardrails filter and constrain content and safety, not enable action execution. Option E is incorrect because a Knowledge Base stores and retrieves static documents via RAG for informational responses, not for performing actions against APIs or databases.

Answer

A prompt flow to manage multiple prompts

Answer

Bedrock Guardrails to filter inappropriate actions

Answer

A Bedrock Knowledge Base to store static documents

Question 15

A data scientist is fine-tuning a foundation model on Amazon Bedrock for a custom summarization task. Which THREE practices should they follow to optimize the fine-tuning process?

Accepted Answer

Start with a base model that is already strong in the domain.. Option A is correct because selecting a base model already strong in the target domain gives the fine-tuning process a better starting point, reducing the amount of task-specific data and compute needed to reach high summarization quality. Option C is correct because a representative dataset that reflects the target task ensures the model learns the desired summarization style, domain vocabulary, and input-output distribution, which directly improves fine-tuning effectiveness. Option D is correct because monitoring training loss and validation loss lets the data scientist detect overfitting early and apply mitigations such as early stopping, regularization, or more data, keeping generalization strong. Option B is not correct because leaving hyperparameters at defaults ignores task-specific tuning of values like learning rate, batch size, and epoch count, which often materially affects fine-tuning results. Option E is not correct because training for as many epochs as possible typically causes overfitting, degrading validation performance rather than optimizing the process.

Answer

Use the default hyperparameters without tuning.

Answer

Train for as many epochs as possible.