Courseiva

AWS Certified AI Practitioner AIF-C01 (AIF-C01) — Questions 676–750

1000 questions total · 14pages · All types, answers revealed

Take a mock examExam hub
Page 9

Page 10 of 14

Page 11
676
MCQmedium

A team has created a knowledge base in Amazon Bedrock for a Q&A application. After updating the source documents, they notice that the model still returns old information. What is the MOST likely cause?

A.The chunking strategy is incorrect
B.The foundation model has a limited context window
C.The guardrails are filtering out the new information
D.The knowledge base has not been resynchronized after updating the documents
AnswerD

Syncing is required to reflect changes in the vector store.

Why this answer

The vector store must be resynchronized after document updates. Chunking changes, model selection, and guardrails would not cause stale data if the vector store is not refreshed.

Full explanation →
677
Multi-Selecthard

A company is using Amazon SageMaker to manage the lifecycle of their machine learning models. They need to implement a governance framework that includes model versioning, monitoring for drift, and decommissioning of outdated models. Which THREE AWS services or features should they use together to meet these requirements? (Select THREE.)

Select 3 answers
A.AWS CloudTrail
B.SageMaker Pipelines
C.SageMaker Model Registry
D.SageMaker Role Manager
E.SageMaker Model Monitor
AnswersB, C, E

Pipelines can automate the decommissioning of outdated models by triggering endpoints to update or delete.

Why this answer

SageMaker Model Registry handles versioning, SageMaker Model Monitor detects drift, and SageMaker Pipelines can orchestrate decommissioning workflows. Together they cover the governance lifecycle.

Full explanation →
678
MCQhard

A manufacturing company is deploying IoT sensors to monitor equipment performance. The sensors generate continuous unlabeled time-series data with thousands of dimensions. The goal is to detect anomalies indicating potential failures in real time. The data science team has experience with unsupervised learning and wants to use a SageMaker built-in algorithm that can handle high-dimensional data and identify outliers. They also need to reduce the number of dimensions to improve training speed without losing important information. Which approach should they take?

A.Use Amazon SageMaker Linear Learner algorithm
B.Use Amazon SageMaker Random Cut Forest algorithm
C.Use Amazon SageMaker Image Classification algorithm
D.Use Amazon SageMaker Object Detection algorithm
AnswerB

Random Cut Forest is an unsupervised anomaly detection algorithm suited for high-dimensional data.

Why this answer

Amazon SageMaker Random Cut Forest (RCF) is a built-in unsupervised algorithm specifically designed for anomaly detection on high-dimensional time-series data. It works by constructing an ensemble of random trees to isolate outliers, making it ideal for the unlabeled, continuous sensor data described. Additionally, RCF inherently handles high-dimensional data without requiring explicit dimensionality reduction, as it randomly samples features at each split, effectively reducing the impact of irrelevant dimensions while preserving anomaly detection accuracy.

Exam trap

The trap here is that candidates may confuse Random Cut Forest with a dimensionality reduction technique like PCA, but RCF does not reduce dimensions—it randomly samples features per tree to handle high-dimensional data without explicit reduction, while still identifying outliers effectively.

How to eliminate wrong answers

Option A is wrong because Amazon SageMaker Linear Learner is a supervised algorithm used for regression or binary classification, requiring labeled data, and it does not natively perform anomaly detection or dimensionality reduction. Option C is wrong because Amazon SageMaker Image Classification is a supervised algorithm designed for classifying images, not for unsupervised anomaly detection on time-series data. Option D is wrong because Amazon SageMaker Object Detection is a supervised algorithm for identifying and localizing objects within images, which is irrelevant to unlabeled time-series sensor data.

Full explanation →
679
Multi-Selecteasy

A company is using Amazon Rekognition to detect objects in images stored in an S3 bucket. The company needs to ensure that the Rekognition service has permission to read images from the S3 bucket. Which TWO methods can achieve this? (Choose TWO.)

Select 2 answers
A.Attach an IAM role to the Lambda function that calls Rekognition, with an IAM policy granting s3:GetObject on the bucket
B.Create an IAM role for Rekognition to assume and attach it to the Rekognition service
C.Configure the S3 bucket with a bucket ACL that grants read access to the Rekognition service
D.Attach a resource-based policy to the Rekognition collection allowing S3 access
E.Add a bucket policy that grants the Rekognition service principal (rekognition.amazonaws.com) permission to read objects
AnswersA, E

The IAM role provides permissions to the Lambda function to read from S3.

Why this answer

Option A is correct because when a Lambda function calls Amazon Rekognition, the function needs an IAM role with a policy that grants s3:GetObject permission on the source bucket. This allows the Lambda function to read the images and pass them to Rekognition for analysis. Option E is correct because you can attach a bucket policy that explicitly grants the Rekognition service principal (rekognition.amazonaws.com) permission to read objects, enabling Rekognition to directly access the S3 bucket when invoked.

Exam trap

The trap here is that candidates often confuse which AWS services can assume IAM roles (like Lambda or EC2) versus those that require resource-based policies (like Rekognition or S3), leading them to incorrectly select option B.

Full explanation →
680
MCQmedium

A developer needs to choose a model on Amazon Bedrock for a text summarization task. The summaries must be accurate and concise, and the input documents are up to 5,000 tokens. Which model selection criteria should be prioritized?

Answer options not yet available.

Why this answer

For summarization, capability and context window are key. The model must handle the input length and produce high-quality summaries.

Full explanation →
681
MCQeasy

A company uses a generative AI model to create marketing copy. They want to ensure that customers know the content is AI-generated. Which practice directly addresses this transparency requirement?

A.Implement AWS CloudTrail to log all model inference calls
B.Add a watermark or disclaimer stating 'This content was generated by AI'
C.Store all generated content in Amazon S3 with versioning enabled
D.Use a more powerful model to generate more natural-sounding text
AnswerB

Explicit disclosure that content is AI-generated meets transparency requirements under emerging regulations and responsible AI guidelines.

Why this answer

Transparency in AI-generated content involves clearly disclosing to users that the content was produced by AI. This can be done through disclaimers, labeling, or other communication methods.

Full explanation →
682
Multi-Selectmedium

A company is deploying a generative AI application using Amazon Bedrock and needs to optimize costs for a high-volume, latency-tolerant workload. Which TWO strategies should they implement? (Select TWO.)

Select 2 answers
A.Use Batch Inference for asynchronous processing
B.Deploy a large model and fine-tune it
C.Use a smaller, more efficient foundation model
D.Enable Provisioned Throughput for guaranteed capacity
E.Implement model caching to avoid redundant inferences
AnswersA, C

Batch inference is cheaper per request and suitable for non-real-time workloads.

Why this answer

Using a smaller model reduces per-request cost, and batch inference processes large volumes asynchronously at lower cost. Provisioned Throughput increases cost; caching helps but is not a primary cost optimization for high-volume; fine-tuning adds cost.

Full explanation →
683
MCQhard

A company wants to use Amazon Bedrock to generate personalized marketing emails. They have thousands of customer profiles with demographic data. To generate tailored content efficiently, the application must dynamically insert customer-specific information into prompts. Which prompt management technique is BEST suited for this?

A.Using a larger foundation model to understand the entire customer base
B.Prompt flows with variables
C.Fine-tuning the model on all customer profiles
D.Prompt versioning
AnswerB

Prompt flows let you define templates with placeholders that are populated with customer data for each request.

Why this answer

Prompt flows allow you to create templates with variables that are replaced at runtime, enabling dynamic prompt generation for each customer.

Full explanation →
684
MCQmedium

A company wants to use Amazon Bedrock to generate product descriptions in multiple languages. They need the model to produce English, Spanish, and French descriptions with equal quality. Which model selection BEST meets this requirement?

A.Cohere Command
B.Amazon Titan Text Express
C.Anthropic Claude
D.Meta Llama 2
AnswerC

Claude offers robust multilingual support.

Why this answer

Anthropic Claude is known for strong multilingual capabilities. Amazon Titan models primarily excel in English. Mistral also supports multiple languages but Claude is generally considered top-tier for multilingual tasks.

Full explanation →
685
MCQmedium

A company is evaluating the performance of a summarization model using Amazon Bedrock model evaluation. They want an automated metric that measures how well the generated summary captures the meaning of the reference summary. Which metric is MOST suitable?

A.BLEU
B.ROUGE-L
C.Accuracy
D.BERTScore
AnswerD

BERTScore uses contextual embeddings to capture semantic similarity.

Why this answer

BERTScore compares embeddings of generated and reference summaries to measure semantic similarity, making it suitable for summarization. ROUGE and BLEU are n-gram based and less sensitive to meaning.

Full explanation →
686
MCQmedium

A data scientist is using Amazon SageMaker Clarify to analyze a model and discovers that the model treats two different demographic groups differently when they should have similar outcomes. The data scientist wants to quantify this difference using a metric that compares the proportion of positive outcomes for each group. Which metric should be used?

A.Disparate impact
B.Demographic parity
C.Equalized odds
D.SHAP values
AnswerB

Demographic parity (statistical parity) measures the difference in positive outcome rates between groups.

Why this answer

Demographic parity measures whether the proportion of positive outcomes is equal across groups. It is also known as statistical parity.

Full explanation →
687
MCQmedium

A team is building a classifier to detect fraudulent transactions. The dataset has 99.9% legitimate transactions and 0.1% fraudulent. Which evaluation metric is most appropriate?

A.Mean Absolute Error (MAE)
B.Root Mean Squared Error (RMSE)
C.AUC-ROC
D.Accuracy
AnswerC

AUC-ROC is robust to class imbalance and evaluates the model's ability to distinguish classes.

Why this answer

Accuracy is misleading for imbalanced data (99.9% if always predict legitimate). F1 score or AUC-ROC better captures performance on minority class. AUC-ROC is insensitive to class imbalance and measures ranking ability.

Full explanation →
688
MCQmedium

Refer to the exhibit. A data scientist created this endpoint config for a foundation model in Amazon SageMaker. However, the endpoint fails to scale under load. What is the most likely reason?

A.Missing AutoScaling configuration
B.Variant weight is 1.0
C.Instance type is too small
D.InitialInstanceCount is 1
AnswerA

Auto scaling policy is required to add instances under load.

Why this answer

The endpoint fails to scale under load because the endpoint configuration shown lacks an AutoScaling policy. Without AutoScaling, SageMaker will not automatically adjust the number of instances based on traffic, so even if the initial instance count is 1, the endpoint cannot add more instances to handle increased load. AutoScaling must be explicitly configured via Application Auto Scaling to define scaling policies and target tracking metrics.

Exam trap

AWS often tests the misconception that setting a higher InitialInstanceCount or choosing a larger instance type alone enables scaling, when in fact AutoScaling must be explicitly configured as a separate step.

How to eliminate wrong answers

Option B is wrong because a variant weight of 1.0 is the default and does not prevent scaling; it simply means all traffic is routed to that variant. Option C is wrong because the instance type being 'too small' would cause performance issues or throttling, but it does not prevent the endpoint from scaling out; scaling is controlled by AutoScaling, not instance size. Option D is wrong because an InitialInstanceCount of 1 is a valid starting point; the endpoint can still scale out if AutoScaling is configured, so a single initial instance does not inherently block scaling.

Full explanation →
689
MCQmedium

A company is training a deep learning model on Amazon SageMaker using a large dataset stored in S3. Training jobs are frequently failing with 'OutOfMemoryError'. The training algorithm uses PyTorch. How should the data scientist solve this without reducing model accuracy?

A.Use SageMaker Pipe mode for data ingestion
B.Reduce the number of layers in the model
C.Increase the batch size
D.Use a smaller instance type with less memory
AnswerA

Pipe mode streams data directly, reducing memory footprint and preventing OutOfMemoryError.

Why this answer

SageMaker Pipe mode streams training data directly from S3 into the algorithm without first downloading it to the local disk, which drastically reduces memory consumption. This allows the model to handle large datasets that would otherwise cause an OutOfMemoryError when using the default File mode, all while preserving the original model architecture and accuracy.

Exam trap

The AIF-C01 exam often tests the misconception that reducing model complexity or instance size is the only way to fix memory errors, when in fact data ingestion mode changes (like Pipe mode) can resolve the issue without sacrificing accuracy or performance.

How to eliminate wrong answers

Option B is wrong because reducing the number of layers in the model would decrease model capacity and likely reduce accuracy, which violates the requirement to not reduce model accuracy. Option C is wrong because increasing the batch size would increase memory usage per training step, exacerbating the OutOfMemoryError rather than solving it. Option D is wrong because using a smaller instance type with less memory would make the memory problem worse, not better, and would likely lead to even more frequent failures.

Full explanation →
690
MCQmedium

A company is using Amazon Bedrock to summarize long documents. They notice that the summary sometimes omits key details. What is the most likely cause?

A.The model is overfitted
B.The prompt lacks examples
C.The model's context window is too small
D.The temperature parameter is too high
AnswerC

A small context window truncates the input document, causing the model to miss key details.

Why this answer

Option A, the model's context window is too small, causes the model to only see part of the document, resulting in omitted details. Option B (temperature too high) increases randomness, not omission. Option C (lack of examples) may affect quality but not omission due to length.

Option D (overfitting) would affect performance on new data, not specifically omission of details.

Full explanation →
691
MCQmedium

A large e-commerce company uses a recommendation system to suggest products to customers. Recently, a data scientist noticed that the model's recommendations for high-value luxury items are predominantly shown to users in affluent zip codes, while users in less affluent areas rarely see these items, even if they have searched for them. The company is concerned about fairness and wants to ensure all customers have equal access to recommendations regardless of location. The current model uses collaborative filtering on historical purchase data. The team needs to modify the system without sacrificing overall recommendation accuracy. Which action best addresses the fairness concern while maintaining performance?

A.Randomly show luxury recommendations to a subset of users regardless of their behavior
B.Remove zip code and any income-correlated features from the training data
C.Add more training data from less affluent areas to balance the dataset
D.Implement a separate recommendation pipeline for luxury items based only on search history
AnswerB

Removing biased features eliminates the source of unfairness in recommendations.

Why this answer

The most effective approach is to ensure the model does not use zip code or any feature correlated with income as a direct or indirect input. This removes the proxy for socioeconomic status. Simply equalizing recommendation frequency artificially may hurt relevance.

Personalizing based on search history is already being done but zip code bias remains. Adding more training data may not help if the bias is in the features.

Full explanation →
692
Multi-Selecteasy

Which TWO factors are MOST important when selecting a foundation model for a text summarization task? (Choose two.)

Select 2 answers
A.Maximum output length (max tokens)
B.Model creation date
C.Model training cost
D.Maximum input length (context window)
E.Image support
AnswersA, D

Determines the length of the summary.

Why this answer

Options A and C are correct. The maximum input length (context window) determines how much text the model can process at once. The output length (max tokens) affects the summary detail.

Options B (training cost) is not a selection factor for pre-trained models. D (image support) irrelevant. E (model creation date) is not a primary factor.

Full explanation →
693
MCQhard

A company needs to generate high-quality images from text descriptions for a marketing campaign. They need to ensure the images are photorealistic and that the model can generate variations of a given image. Which type of model should they use?

Answer options not yet available.

Why this answer

Diffusion models (e.g., Stable Diffusion, Amazon Titan Image Generator) are the state-of-the-art for text-to-image generation and can produce photorealistic results with variations.

Full explanation →
694
MCQhard

During a SageMaker training job, the data scientist observes that the loss is not decreasing after the initial few epochs. The model is a deep neural network with ReLU activations. Which hyperparameter adjustment is most likely to help?

A.Reduce the learning rate
B.Increase the number of epochs
C.Increase the learning rate
D.Decrease the batch size
AnswerA

A lower learning rate can allow the optimizer to find a better minimum.

Why this answer

When loss plateaus after a few epochs with ReLU activations, the model is likely stuck in a region where gradients are small (e.g., near a local minimum or plateau). Reducing the learning rate allows the optimizer to take smaller steps, which can help it navigate out of flat regions and continue decreasing the loss. This is a standard technique to improve convergence when training stalls.

Exam trap

The AIF-C01 exam often tests the misconception that increasing the learning rate will accelerate convergence, but in plateau scenarios it actually causes divergence or oscillation, making the reduction of learning rate the correct adjustment.

How to eliminate wrong answers

Option B is wrong because increasing the number of epochs does not address the underlying issue of the optimizer being unable to escape a plateau; it would simply continue training with no improvement. Option C is wrong because increasing the learning rate would likely cause the optimizer to overshoot the minimum or oscillate, potentially worsening the loss plateau. Option D is wrong because decreasing the batch size introduces more noise into gradient estimates, which can destabilize training and does not directly help when the loss is stuck on a plateau.

Full explanation →
695
MCQmedium

A company is using Amazon SageMaker JumpStart to deploy a pre-trained text generation model. After deployment, the model produces slow inference responses. Which action is most likely to improve inference latency?

A.Quantize the model weights to FP16 or INT8.
B.Deploy the model on a more powerful instance type with higher GPU memory.
C.Fine-tune the model on a smaller dataset.
D.Increase the batch size for inference requests.
AnswerB

More compute resources reduce inference time per request.

Why this answer

Option B is correct because deploying the model on a more powerful instance type with higher GPU memory directly addresses the computational bottleneck causing slow inference. A larger GPU provides more CUDA cores and memory bandwidth, enabling faster matrix operations and reducing the time per forward pass for the pre-trained text generation model.

Exam trap

The AIF-C01 exam often tests the misconception that model optimization techniques like quantization always improve latency without trade-offs, but the most direct and reliable method for reducing inference latency is upgrading to a more powerful instance type with higher GPU memory.

How to eliminate wrong answers

Option A is wrong because quantizing model weights to FP16 or INT8 reduces model size and can improve latency, but it may degrade output quality and is not the most direct or guaranteed fix for slow inference; the question asks for the action most likely to improve latency, and upgrading hardware is more reliable. Option C is wrong because fine-tuning on a smaller dataset adjusts the model for a specific task but does not inherently speed up inference; it may even increase latency if the fine-tuned model is larger or uses more complex attention patterns. Option D is wrong because increasing batch size for inference requests typically increases throughput (requests per second) but can increase per-request latency due to longer queue times and higher memory usage, making it counterproductive for reducing individual response time.

Full explanation →
696
MCQeasy

Which AWS service can be used to convert a recorded speech file into text for further analysis?

A.Amazon Polly
B.Amazon Translate
C.Amazon Rekognition
D.Amazon Transcribe
AnswerD

Transcribe converts audio to text.

Why this answer

Amazon Transcribe is an automatic speech recognition (ASR) service that converts audio to text. Polly does the opposite (text to speech). Rekognition is for images/video, and Translate is for language translation.

Full explanation →
697
MCQmedium

A company is using Amazon Bedrock to generate text summaries of customer emails. The compliance team requires that any email containing a Social Security Number (SSN) must be blocked from being sent to the model for summarization. Which Bedrock Guardrail configuration should be used?

A.Configure a topic restriction to block 'PII' topics
B.Enable model invocation logging and manually review all inputs
C.Add a word filter with a list of common SSN patterns
D.Use a content policy with a PII filter set to 'Deny' for SSN
AnswerD

Bedrock Guardrails include managed PII filters that can deny input containing specific PII types like SSNs.

Why this answer

Bedrock Guardrails can be configured to filter content before it is sent to the model. PII redaction with a deny action will block the input if SSNs are detected. The guardrail is applied at invocation time.

Full explanation →
698
MCQeasy

A company wants to extract text and data from scanned PDF invoices for automated processing. Which AWS service is MOST appropriate for this task?

A.Amazon Rekognition
B.Amazon Transcribe
C.Amazon Comprehend
D.Amazon Textract
AnswerD

Textract is built for extracting text and data from scanned documents, including invoices.

Why this answer

Amazon Textract is specifically designed for extracting text, handwriting, and data from scanned documents. It supports forms and tables, making it ideal for invoices. The other services are for other modalities: Rekognition for images/video, Comprehend for NLP, and Transcribe for audio.

Full explanation →
699
MCQhard

An enterprise is building a RAG solution with Amazon Bedrock and needs to ensure that the retrieved documents are from authorised sources only. They also must prevent the model from generating responses that contain personally identifiable information (PII). Which two Bedrock features combined address these requirements?

A.Amazon Bedrock Knowledge Bases and Bedrock Guardrails
B.Amazon Bedrock Model Evaluation and Bedrock Studio
C.Amazon Bedrock Agents and Bedrock Guardrails
D.Amazon Bedrock Playground and prompt engineering
AnswerA

Knowledge Bases restrict retrieval to indexed documents, Guardrails handle PII detection.

Why this answer

Guardrails with PII detection can mask or block PII. Knowledge Bases define the authorised data sources for retrieval. Agents are not needed for source control.

Full explanation →
700
MCQmedium

A bank wants to use Amazon Augmented AI (A2I) to review high-value loan applications that require human judgment. Which workflow best implements human-in-the-loop review for these predictions?

A.Set up an A2I workflow with a confidence threshold, so low-confidence predictions are sent to human reviewers
B.Route all loan applications to human reviewers for approval
C.Use Amazon Mechanical Turk to review all predictions in real time
D.Configure a private workforce in Amazon SageMaker Ground Truth
AnswerA

A2I allows setting conditions like a confidence threshold to trigger human review, balancing automation and human oversight.

Why this answer

Amazon A2I enables human review of low-confidence predictions or specific conditions. The best practice is to set a confidence threshold; predictions below that threshold are sent to human reviewers.

Full explanation →
701
MCQmedium

A financial services company deploys a generative AI chatbot for customer support. They want to prevent the chatbot from generating harmful or misleading information. Which AWS service can help monitor and filter responses?

A.Amazon GuardDuty
B.Amazon Augmented AI (A2I) with human review
C.AWS WAF
D.Amazon Comprehend
AnswerB

A2I can route responses for human review to ensure safety.

Why this answer

Option A is correct: Amazon Augmented AI (A2I) with human review can monitor and filter responses for sensitive content. Option B is wrong: Amazon Comprehend is for understanding, not filtering. Option C is wrong: AWS WAF is for web traffic.

Option D is wrong: Amazon GuardDuty is for security threats.

Full explanation →
702
Multi-Selectmedium

A machine learning engineer is implementing a RAG system using Amazon Bedrock and a vector database. They need to chunk a large set of PDF documents before embedding. Which THREE considerations are important for chunking strategy? (Select THREE.)

Select 3 answers
A.Overlap between consecutive chunks
B.The method used to split chunks (e.g., by sentence, paragraph, or fixed token count)
C.Chunk size (e.g., number of tokens)
D.The embedding model's maximum input length
E.The metadata associated with each document
AnswersA, B, C

Overlap helps maintain context across chunks.

Why this answer

Chunk size affects retrieval granularity; overlap ensures context continuity; chunking method (e.g., sentence splitting) influences semantic coherence. Document metadata is not part of chunking; embedding model choice is separate from chunking strategy.

Full explanation →
703
Multi-Selectmedium

A data scientist is preparing data for a classification task. Which TWO techniques are commonly used for handling missing values? (Choose two.)

Select 2 answers
A.Label encoding
B.Normalization
C.Imputing with mean
D.Dropping rows with any missing values
E.One-hot encoding
AnswersC, D

Mean imputation replaces missing values with the mean of the column.

Why this answer

Imputing with the mean is a common technique for handling missing values in numerical features because it preserves the overall distribution of the data without reducing the dataset size. This method replaces each missing entry with the arithmetic mean of the non-missing values in that column, which is simple to implement and works well when data is missing completely at random (MCAR).

Exam trap

The AIF-C01 exam often tests the distinction between data preprocessing techniques (e.g., encoding, scaling) and missing value handling, so candidates mistakenly select label encoding or normalization because they are common preprocessing steps, even though they do not address missing data.

Full explanation →
704
MCQmedium

Refer to the exhibit. A SageMaker training job fails with an 'AccessDenied' error when trying to read files from the S3 bucket 'my-training-data'. The IAM role used by the training job has the policy shown. What is the most likely reason for the failure?

A.The bucket policy requires encryption in transit
B.The training job is using the wrong AWS region
C.The policy does not include the s3:PutObject action
D.The policy does not include the s3:ListBucket action
AnswerD

Without ListBucket, SageMaker cannot list the contents of the bucket to verify object existence.

Why this answer

The IAM policy grants s3:GetObject but not s3:ListBucket. When a SageMaker training job reads files from S3, the SageMaker SDK or framework (e.g., TensorFlow, PyTorch) often performs a ListBucket call first to enumerate objects in the prefix. Without s3:ListBucket, the SDK cannot discover the files, resulting in an AccessDenied error even though GetObject is allowed.

Exam trap

AWS often tests the misconception that only s3:GetObject is needed for reading from S3, but the SDK's underlying ListBucket call is required for object discovery, especially when using prefixes or manifest files.

How to eliminate wrong answers

Option A is wrong because the error is an IAM AccessDenied, not a bucket policy condition failure; encryption in transit would cause a different error (e.g., 'The request was denied because of a condition in the bucket policy'). Option B is wrong because the training job and S3 bucket must be in the same region for SageMaker to access the data, but the error message would be 'BucketRegionError' or a timeout, not AccessDenied. Option C is wrong because s3:PutObject is not needed for reading files; the training job only requires read permissions (GetObject and ListBucket) to fetch training data.

Full explanation →
705
Multi-Selecteasy

Which TWO statements about the bias-variance tradeoff are correct? (Choose TWO.)

Select 2 answers
A.High bias typically leads to overfitting
B.High variance models are insensitive to changes in training data
C.High bias typically leads to underfitting
D.Increasing model complexity typically decreases variance
E.High variance typically leads to overfitting
AnswersC, E

High bias means the model is too simple and underfits the data.

Why this answer

High bias leads to underfitting (simple model), high variance leads to overfitting (complex model). The goal is to balance them. Option B is correct: high bias often underfits.

Option D is correct: high variance often overfits. Option A is false because high bias causes underfit, not overfit. Option C is false because models with high variance are sensitive to training data.

Option E is false because increasing model complexity increases variance.

Full explanation →
706
MCQhard

A legal firm uses Amazon Bedrock to generate contract summaries. They want to evaluate the quality of summaries against human-written reference summaries. The evaluation should capture both the overlap of n-grams and the semantic similarity. Which combination of automated metrics is MOST appropriate?

A.Exact match and F1 score
B.ROUGE and BLEU
C.ROUGE and BERTScore
D.BLEU and BERTScore
AnswerC

ROUGE captures n-gram overlap for summarization; BERTScore captures semantic similarity via embeddings. Together they cover both lexical and semantic quality.

Why this answer

ROUGE measures n-gram overlap (recall-oriented) for summarization, while BERTScore uses contextual embeddings to capture semantic similarity. BLEU is for translation. Combining ROUGE and BERTScore gives both lexical and semantic evaluation.

Full explanation →
707
MCQmedium

A financial services firm needs to ensure that all calls to Amazon Bedrock APIs are logged for audit purposes. Which AWS service should they enable to capture API calls?

A.AWS CloudTrail
B.Amazon S3 server access logs
C.Amazon CloudWatch Logs
D.AWS Config
AnswerA

CloudTrail records all AWS API calls, including Bedrock, for auditing and compliance.

Why this answer

AWS CloudTrail records API activity in AWS accounts, including Bedrock API calls, providing audit logs.

Full explanation →
708
Multi-Selectmedium

A data science team uses Amazon SageMaker to train models. To comply with SOC 2, they must ensure that access to training data is logged, that the data is encrypted at rest, and that model training jobs are isolated from each other. Which THREE actions should they take? (Choose three.)

Select 3 answers
A.Enable Amazon Inspector to scan training instances for vulnerabilities.
B.Enable server-side encryption on the S3 bucket containing training data using SSE-KMS.
C.Use SageMaker Debugger to monitor training jobs.
D.Enable AWS CloudTrail to capture SageMaker API calls.
E.Use SageMaker VPC mode to launch training jobs in a private subnet.
AnswersB, D, E

SSE-KMS encrypts data at rest.

Why this answer

Option B is correct because enabling server-side encryption on the S3 bucket containing training data using SSE-KMS ensures data at rest is encrypted, which is a direct requirement for SOC 2 compliance. SSE-KMS provides envelope encryption with a customer-managed AWS KMS key, allowing fine-grained access control and audit trails for the encryption keys.

Exam trap

The trap here is that candidates may confuse Amazon Inspector with a logging or encryption service, or think SageMaker Debugger provides security logging, when in fact Inspector only scans for vulnerabilities and Debugger only monitors model training metrics.

Full explanation →
709
MCQmedium

During model training, a data scientist notices that the model performs very well on the training data but poorly on the test data. The scientist suspects high variance. Which technique is MOST likely to reduce the variance and improve test performance?

A.Lower the learning rate
B.Increase the number of features
C.Apply regularization (e.g., L1 or L2)
D.Decrease the amount of training data
AnswerC

Regularization penalizes large weights, simplifying the model and reducing overfitting.

Why this answer

High variance (overfitting) means the model is too complex. Regularization adds a penalty for large coefficients, reducing overfitting. Decreasing training data would increase variance, increasing model complexity would worsen it, and lowering the learning rate may not address variance directly.

Full explanation →
710
Multi-Selecthard

A company is deploying a generative AI application using Amazon Bedrock and needs to optimize costs. They expect variable traffic with occasional high spikes. Which TWO strategies would help reduce costs while maintaining performance?

Select 2 answers
A.Enable model caching to serve repeated prompts without re-inference
B.Use on-demand inference with batch processing for non-real-time requests
C.Always use the largest available foundation model for quality
D.Use provisioned throughput to handle spikes
E.Fine-tune a larger model for better accuracy
AnswersA, B

Caching avoids repeated computation for identical requests.

Why this answer

Model caching reduces cost for repeated queries, and batch inference is cheaper for non-real-time workloads. Provisioned throughput is expensive and not cost-optimal for variable traffic. Fine-tuning and using the largest model increase costs.

Full explanation →
711
MCQmedium

A company wants to build a customer service chatbot that answers questions about their internal policy documents. The documents are updated monthly, and the team cannot afford to retrain a model each time. Which approach is MOST appropriate?

A.Train a custom model from scratch on the policy documents each month
B.Use a larger foundation model with a longer context window and paste all documents into each prompt
C.Use Retrieval-Augmented Generation (RAG) with the policy documents indexed in a vector store
D.Fine-tune a base LLM on the policy documents monthly
AnswerC

RAG retrieves relevant document chunks at query time, ensuring the chatbot always answers from the latest uploaded documents without any model retraining.

Why this answer

RAG (Retrieval-Augmented Generation) allows the LLM to retrieve relevant document sections at inference time, so knowledge stays current without retraining. The other options either require expensive retraining for each update or lack document grounding.

Full explanation →
712
MCQeasy

A data scientist wants to fine-tune a foundation model on a specific domain dataset using Amazon SageMaker. Which built-in SageMaker feature can simplify the training process?

A.SageMaker Neo
B.SageMaker Canvas
C.SageMaker JumpStart
D.SageMaker Ground Truth
AnswerC

JumpStart offers one-click fine-tuning for many foundation models, with built-in notebooks and scripts.

Why this answer

Option A, SageMaker JumpStart, provides pre-trained foundation models and built-in fine-tuning scripts, simplifying the process. Option B (Ground Truth) is for data labeling. Option C (Neo) is for model optimization.

Option D (Canvas) is a no-code ML tool for business analysts.

Full explanation →
713
Multi-Selectmedium

A healthcare organization is deploying an AI system to assist in diagnosing diseases from medical images. They need to ensure the system is robust, safe, and subject to human oversight. Which TWO actions align with responsible AI guidelines? (Select TWO.)

Select 2 answers
A.Monitor model performance over time to detect data drift
B.Deploy the model directly to production without testing
C.Use Amazon Augmented AI (A2I) to set up human review for uncertain predictions
D.Permanently disable human review to speed up diagnoses
E.Use only one source of training data
AnswersA, C

Data drift can degrade model accuracy; monitoring ensures robustness and triggers retraining when needed.

Why this answer

Monitoring for data drift and integrating human review for high-risk decisions are key practices for robustness, safety, and controllability. The other options are either insufficient or irrelevant.

Full explanation →
714
MCQeasy

A developer is using Amazon Bedrock to create a chatbot. They want to ensure the bot does not generate toxic or offensive content. Which feature should they enable?

A.Use careful prompt engineering to avoid toxic responses.
B.Fine-tune the model on a dataset of safe responses.
C.Enable content filtering on the Bedrock model.
D.Implement external response validation using a third-party API.
AnswerC

Content filtering provides automated detection and blocking of inappropriate content.

Why this answer

Option B is correct because Bedrock offers content filtering to detect and block harmful content. Option A is wrong because fine-tuning may not fully filter toxic content. Option C is wrong because prompt engineering alone is not enough.

Option D is wrong because response validation is not a built-in feature of Bedrock.

Full explanation →
715
MCQmedium

A company is developing an AI system that transcribes medical consultations. To ensure privacy and security, they need to implement controls that protect patient health information (PHI). Which AWS service can help anonymize data before it is used for model training?

A.Amazon SageMaker Ground Truth
B.AWS Lake Formation
C.Amazon Rekognition
D.Amazon Comprehend Medical
AnswerD

Comprehend Medical can detect PHI entities, enabling redaction or anonymization.

Why this answer

Amazon Comprehend Medical can detect and extract PHI, and AWS Glue can assist in data preparation, but for direct anonymization, services like Amazon Macie or custom solutions may be used. However, among the options, Amazon Comprehend Medical is most directly related to identifying PHI in medical text.

Full explanation →
716
MCQhard

A data scientist is building a RAG application using Amazon Bedrock Knowledge Bases. They want to ensure that only the most semantically relevant documents are retrieved for each query. Which embedding model characteristic is MOST important for this requirement?

A.Large training dataset size
B.High embedding dimension
C.Long context window
D.Low inference latency
AnswerB

Correct. Higher dimensions can represent more nuanced semantics, improving relevance.

Why this answer

Semantic relevance depends on the quality of embeddings. High-dimensional embeddings (e.g., 1024-dim) can capture finer semantic nuances compared to lower dimensions, leading to better retrieval accuracy. Training data diversity, inference latency, and context window length are secondary factors.

Full explanation →
717
MCQhard

A financial institution uses Amazon Bedrock to generate investment summaries. They must prevent the model from discussing prohibited topics like insider trading. Which Bedrock feature should be applied to the model invocation?

A.PII detection filter
B.Content filtering
C.Topic denial policy
D.Grounding check
AnswerC

Correct. Topic denial policies in Guardrails allow you to block entire topics.

Why this answer

Bedrock Guardrails allow topic denial — you can define a list of prohibited topics (e.g., insider trading) and the guardrail will block the model from generating responses about them. The other options address different concerns (PII, content filtering, grounding).

Full explanation →
718
MCQeasy

A company wants to track API calls made to Amazon SageMaker for audit purposes. Which AWS service should they enable?

A.AWS CloudTrail
B.Amazon Macie
C.AWS Config
D.Amazon CloudWatch Logs
AnswerA

CloudTrail records all API calls for auditing and compliance.

Why this answer

AWS CloudTrail is the correct service because it records API activity across AWS services, including Amazon SageMaker. By enabling CloudTrail, the company can capture all SageMaker API calls (e.g., CreateModel, InvokeEndpoint) for audit, compliance, and security analysis. CloudTrail logs provide details such as the identity of the caller, the time of the call, and the request parameters, which are essential for auditing.

Exam trap

The trap here is that candidates often confuse AWS CloudTrail (for API auditing) with Amazon CloudWatch Logs (for log monitoring), mistakenly thinking CloudWatch Logs is the primary service for tracking API calls, but CloudTrail is the dedicated service for recording API activity across AWS.

How to eliminate wrong answers

Option B (Amazon Macie) is wrong because Macie is a data security service that uses machine learning to discover, classify, and protect sensitive data in Amazon S3, not to track API calls. Option C (AWS Config) is wrong because Config evaluates and records resource configuration changes (e.g., SageMaker endpoint configuration), not API call activity. Option D (Amazon CloudWatch Logs) is wrong because CloudWatch Logs is for monitoring, storing, and accessing log files from applications and AWS services, but it does not natively capture API calls; it can ingest CloudTrail logs but is not the primary service for API auditing.

Full explanation →
719
MCQmedium

A developer is trying to invoke the Claude v2 model in Amazon Bedrock from a Lambda function. The Lambda function's IAM role has the following policy attached: { "Version": "2012-10-17", "Statement": [ { "Effect": "Allow", "Action": "bedrock:InvokeModel", "Resource": "*" } ] } When the Lambda function runs, it receives the error shown in the exhibit. Which additional step is most likely needed to resolve this issue?

A.Change the AWS region to one where Claude v2 is available.
B.Use a different model ID such as 'anthropic.claude-v1'.
C.Request access to the Anthropic Claude model through the Amazon Bedrock console.
D.Add a condition to the IAM policy to specify the model ARN.
AnswerC

Model access must be explicitly granted per model even with IAM permissions.

Why this answer

In Amazon Bedrock, even with IAM permissions allowing access to all models, you must also request access to specific foundation models through the AWS console. The AccessDeniedException here indicates the model is not enabled for the account. Option A is correct.

Option B is incorrect because the policy already allows all models. Option C is incorrect because the region is irrelevant to this error. Option D is incorrect because the API call is correct.

Full explanation →
720
MCQeasy

A company wants to build a generative AI application that generates personalized marketing emails based on customer data. They have a small dataset of past emails. Which AWS service should they use to fine-tune a foundation model with their data?

A.Amazon SageMaker
B.Amazon Comprehend
C.AWS Lambda
D.Amazon Bedrock
AnswerA

SageMaker with JumpStart allows fine-tuning of foundation models using custom datasets and provides managed training infrastructure.

Why this answer

Amazon SageMaker provides a managed environment for training and fine-tuning models, including foundation models via JumpStart. Bedrock offers managed APIs but not direct fine-tuning. Lambda is for serverless code, not model training.

Comprehend is for NLP analysis, not text generation.

Full explanation →
721
MCQmedium

A healthcare startup is using Amazon SageMaker to train a model on patient data. They need to ensure that the training data does not contain any personally identifiable information (PII) before being used. Which AWS service can automatically detect and report PII in the data stored in S3?

A.Amazon Rekognition
B.AWS Glue DataBrew
C.Amazon Comprehend Medical
D.Amazon Macie
AnswerD

Macie is the correct service for automated discovery of sensitive data in S3.

Why this answer

Amazon Macie uses machine learning to automatically discover, classify, and protect sensitive data in S3. It can detect PII such as names, addresses, and health information.

Full explanation →
722
MCQhard

A data scientist is evaluating two different foundation models for a summarization task. They want to compare the quality of summaries generated by each model against a set of human-written reference summaries. Which set of metrics is most appropriate for this automated evaluation?

A.Accuracy, precision, recall, and F1-score
B.ROUGE, BLEU, and BERTScore
C.Latency and throughput
D.Mean squared error (MSE) and R-squared
AnswerB

ROUGE and BLEU measure n-gram overlap with references, while BERTScore captures semantic similarity, making them ideal for summarization evaluation.

Why this answer

ROUGE, BLEU, and BERTScore are standard metrics for comparing generated text to reference summaries, covering n-gram overlap and semantic similarity.

Full explanation →
723
MCQeasy

A startup wants to generate product descriptions from a few keywords using a foundation model. They need a fully managed serverless solution that requires no infrastructure setup. Which AWS service should they use?

A.Amazon SageMaker
B.Amazon Comprehend
C.AWS Lambda
D.Amazon Bedrock
AnswerD

Bedrock is a serverless service offering foundation models via API.

Why this answer

Amazon Bedrock is a fully managed serverless service that provides access to foundation models (FMs) from leading AI providers via a simple API, making it ideal for generating product descriptions from keywords without any infrastructure management. It directly supports generative AI tasks like text generation, unlike other AWS services that focus on different ML or NLP capabilities.

Exam trap

The trap here is that candidates may confuse Amazon SageMaker's managed ML capabilities with a serverless generative AI service, overlooking that SageMaker requires explicit infrastructure setup for model hosting, while Bedrock is purpose-built for serverless access to foundation models.

How to eliminate wrong answers

Option A is wrong because Amazon SageMaker is a fully managed machine learning platform that requires setting up training jobs, endpoints, and infrastructure for custom models, not a serverless solution for directly using pre-built foundation models. Option B is wrong because Amazon Comprehend is a natural language processing (NLP) service for tasks like sentiment analysis and entity extraction, not for generative text creation from keywords. Option C is wrong because AWS Lambda is a serverless compute service that runs custom code but does not natively provide access to foundation models; you would need to integrate it with another service like Bedrock to generate descriptions, making it not a standalone solution for this use case.

Full explanation →
724
MCQmedium

A company wants to reduce the cost of running a large number of inference requests for a text classification task. The responses can tolerate a slight delay. Which cost optimization strategy should they implement?

A.Store the training data in a vector store for faster retrieval
B.Use batch inference to process multiple requests asynchronously
C.Use a larger, more accurate model to reduce the number of retries
D.Implement model caching to reuse results for identical prompts
AnswerB

Batch inference groups requests, lowering per-request cost, and is suitable when real-time responses are not required.

Why this answer

Batch inference processes multiple requests together, reducing cost per request at the expense of higher latency. Model caching can help if there are many repeated requests, but batch inference is more general. Switching to a larger model increases cost.

Using a vector store is unrelated.

Full explanation →
725
MCQhard

Refer to the exhibit. A developer deploys this CloudFormation stack but the agent fails to query the knowledge base. What is a likely cause?

A.The KnowledgeBaseId is not passed correctly
B.The agent role does not have permissions to invoke the knowledge base
C.The embedding model is not available in the region
D.The OpenSearch collection type should be SEARCH not VECTORSEARCH
AnswerB

The agent's IAM role must have bedrock:InvokeKnowledgeBase permission.

Why this answer

The correct answer is B because the agent role must have an IAM policy that grants the `bedrock:Retrieve` and `bedrock:RetrieveAndGenerate` permissions on the knowledge base. Without these permissions, the agent cannot invoke the knowledge base, even if the KnowledgeBaseId is correctly passed and the embedding model is available.

Exam trap

AWS often tests the distinction between resource creation permissions and runtime invocation permissions, trapping candidates who assume that a successful stack deployment implies all runtime permissions are correctly configured.

How to eliminate wrong answers

Option A is wrong because if the KnowledgeBaseId were not passed correctly, the stack would likely fail during creation or the agent would receive a different error (e.g., resource not found), not a generic failure to query. Option C is wrong because if the embedding model were not available in the region, the CloudFormation stack itself would fail during creation of the knowledge base, not during a subsequent query. Option D is wrong because the OpenSearch collection type for a knowledge base must be `VECTORSEARCH` to store and query vector embeddings; `SEARCH` is used for full-text search and does not support the vector similarity search required by the knowledge base.

Full explanation →
726
MCQhard

A company using Amazon Bedrock needs to redact personally identifiable information (PII) from user inputs before sending them to the foundation model. Which Bedrock Guardrails component should be configured?

A.Content filters
B.Topic restrictions
C.Word filters
D.Sensitive information filters
AnswerD

Sensitive information filters (PII redaction) detect and redact PII.

Why this answer

PII redaction in Guardrails automatically detects and redacts PII in user inputs or model responses.

Full explanation →
727
MCQmedium

A company is using Amazon Bedrock to build a text-to-SQL application. They want to ensure that the generated SQL queries are valid and safe. Which approach is BEST?

A.Fine-tune the model on a dataset of valid SQL queries
B.Use a separate model to validate the SQL after generation
C.Configure a guardrail to filter and validate the generated SQL
D.Limit the max_tokens to 50 to reduce complexity
AnswerC

Guardrails can enforce rules and reject invalid queries.

Why this answer

Amazon Bedrock guardrails provide a native, configurable mechanism to filter and validate model outputs, including SQL queries, against defined policies such as regex patterns, denied topics, and content filters. This approach directly addresses both validity and safety without requiring additional model training or external validation services, making it the most integrated and efficient solution for ensuring generated SQL is syntactically correct and free of harmful operations like DROP or DELETE.

Exam trap

The AIF-C01 exam often tests the misconception that fine-tuning or output length limits can solve safety and validity requirements, when in fact guardrails are the purpose-built AWS service for content filtering and validation at inference time.

How to eliminate wrong answers

Option A is wrong because fine-tuning on valid SQL queries improves the model's ability to generate syntactically correct SQL but does not guarantee safety; a fine-tuned model can still produce harmful queries (e.g., DROP TABLE) if the training data includes such patterns or if the model generalizes incorrectly. Option B is wrong because using a separate model for validation introduces additional latency, cost, and complexity, and still requires a policy or rule set to define what constitutes 'valid and safe'—which is exactly what Bedrock guardrails already provide natively. Option D is wrong because limiting max_tokens to 50 does not ensure SQL validity or safety; it only truncates output, potentially producing incomplete or syntactically invalid SQL, and does not prevent generation of dangerous commands.

Full explanation →
728
MCQeasy

A developer is testing different prompts for a text generation model on Amazon Bedrock. Which parameter controls the randomness of the model's output?

A.top_p
B.stop_sequences
C.temperature
D.max_tokens
AnswerC

Temperature directly scales the logits before softmax, controlling the randomness of token selection.

Why this answer

Option D is correct because temperature controls the randomness of the model's predictions. Lower values make output more deterministic; higher values increase randomness. Option A (max_tokens) controls output length.

Option B (top_p) is nucleus sampling. Option C (stop_sequences) defines stopping criteria.

Full explanation →
729
MCQhard

An organization is using Amazon Bedrock to power a customer service chatbot. They notice that the chatbot occasionally generates hallucinated information about product specifications. Which strategy should be implemented to reduce hallucinations?

A.Fine-tune the model on a dataset of product specification conversations.
B.Integrate a Retrieval Augmented Generation (RAG) system with the product catalog.
C.Use more detailed prompts with explicit instructions to avoid speculation.
D.Increase the temperature parameter to make outputs more conservative.
AnswerB

RAG provides up-to-date, factual context to the model, reducing hallucinations.

Why this answer

Retrieval Augmented Generation (RAG) grounds the model's responses in authoritative, up-to-date product catalog data, directly reducing hallucinations by ensuring the chatbot references verified facts rather than relying solely on its parametric memory. This is the most effective strategy because it provides a retrieval-based factual foundation that fine-tuning or prompt engineering alone cannot guarantee.

Exam trap

The AIF-C01 exam often tests the misconception that prompt engineering or fine-tuning alone can solve hallucination problems, when in fact they lack the dynamic, verifiable grounding that RAG provides.

How to eliminate wrong answers

Option A is wrong because fine-tuning on product specification conversations may reinforce patterns from the training data but does not prevent the model from generating plausible-sounding but incorrect details when faced with queries outside the fine-tuned distribution; it also cannot dynamically incorporate real-time catalog updates. Option C is wrong because while more detailed prompts can reduce speculation, they do not provide the model with access to external, authoritative data—hallucinations can still occur when the model's internal knowledge is incomplete or outdated. Option D is wrong because increasing the temperature parameter makes outputs more random and creative, not more conservative; decreasing temperature would make outputs more deterministic and less prone to hallucination, but even low temperature cannot eliminate hallucinations without a retrieval mechanism.

Full explanation →
730
MCQhard

A company uses Amazon Bedrock to deploy a foundation model for a real-time chat application. Users report that responses are slow. Which optimization is MOST likely to reduce latency without degrading quality?

A.Use a larger model variant to improve inference speed
B.Increase the temperature to make the model generate faster
C.Enable response streaming using the Converse API or InvokeModelWithResponseStream
D.Switch from a text generation model to an embedding model
AnswerC

Streaming allows the client to display partial results immediately, improving user experience.

Why this answer

Streaming responses return tokens as they are generated, reducing perceived latency. Other options either increase cost or limit model capability.

Full explanation →
731
MCQmedium

A developer needs to reduce costs for a Bedrock application that processes high volumes of similar queries. The queries are repetitive and the model is invoked many times with the same prompt. Which cost optimization technique is MOST suitable?

A.Select a smaller right-sized model
B.Fine-tune the model on the common queries
C.Use batch inference to process multiple queries together
D.Enable model caching to serve repeated prompts from cache
AnswerD

Caching avoids recomputation for identical prompts, reducing latency and cost.

Why this answer

Model caching (prompt caching) stores responses to common prompts, reducing the need to recompute. Batch inference is for asynchronous processing, not real-time. Right-sizing the model helps but does not leverage query repetition.

Fine-tuning is for adapting the model, not cost savings on repetitive queries.

Full explanation →
732
MCQeasy

A company is using Amazon Bedrock to build a generative AI application. The company wants to prevent the model from generating toxic or harmful content while still allowing creative responses. Which feature should the company enable?

A.Amazon Bedrock Guardrails with content filters.
B.AWS Key Management Service (KMS) to encrypt model responses.
C.AWS Identity and Access Management (IAM) policies to restrict model output.
D.Amazon CloudWatch Logs to monitor and block harmful content.
AnswerA

Guardrails provide configurable content filters to block harmful output without overly restricting creativity.

Why this answer

Amazon Bedrock Guardrails with content filters is the correct feature because it allows the company to define and enforce policies that block toxic or harmful content in model inputs and outputs, while still permitting creative responses within safe boundaries. This feature provides configurable thresholds for content categories like hate, insults, and sexual content, enabling precise control over model behavior without restricting overall creativity.

Exam trap

The trap here is that candidates may confuse security services (like KMS for encryption or IAM for access control) with content moderation capabilities, assuming any AWS security service can filter model outputs, when in fact only Bedrock Guardrails provides purpose-built content filters for generative AI.

How to eliminate wrong answers

Option B is wrong because AWS KMS encrypts data at rest and in transit but does not inspect or filter model responses for toxic content; encryption ensures confidentiality, not content safety. Option C is wrong because IAM policies control access to AWS resources and actions (e.g., who can invoke a model) but cannot restrict the actual text output of a model; they are for authorization, not content moderation. Option D is wrong because Amazon CloudWatch Logs can monitor and store logs for analysis but cannot actively block harmful content in real-time; it is a logging and monitoring service, not a content filter.

Full explanation →
733
MCQeasy

A startup is deploying a foundation model on Amazon SageMaker for real-time inference. They notice high latency (over 2 seconds per request). Which action is most likely to reduce latency?

A.Enable auto-scaling on the SageMaker endpoint to handle more concurrent requests.
B.Switch to a smaller, distilled version of the model.
C.Deploy the model on a CPU-based instance instead of GPU.
D.Increase the batch size parameter in the inference request.
AnswerB

Smaller models have fewer parameters, reducing computation time and latency.

Why this answer

Option B is correct because using a smaller, distilled version of the model directly reduces the computational complexity per inference request. Distillation compresses the model by training a smaller student network to mimic a larger teacher model, resulting in fewer parameters and faster forward passes. This is the most direct way to cut latency when the model size is the bottleneck, as it reduces the number of floating-point operations (FLOPs) required per request.

Exam trap

AWS often tests the distinction between latency (time per single request) and throughput (requests per second), so candidates mistakenly choose auto-scaling or batch size increases, which improve throughput but not per-request latency.

How to eliminate wrong answers

Option A is wrong because enabling auto-scaling adds more endpoint instances to handle higher concurrency, but it does not reduce the latency of a single inference request; it only improves throughput under load. Option C is wrong because CPU-based instances are generally slower for deep learning inference than GPU instances, especially for large foundation models, so switching to CPU would increase latency, not reduce it. Option D is wrong because increasing the batch size in the inference request means processing multiple inputs together, which increases the time to first byte for each individual request and does not reduce per-request latency; it is a throughput optimization, not a latency reduction technique.

Full explanation →
734
Multi-Selecteasy

A company needs to build a system that converts text-based customer reviews into audio files for accessibility. Which TWO AWS services should be used? (Choose two.)

Select 1 answer
A.Amazon Transcribe
B.Amazon Translate
C.Amazon Rekognition
D.Amazon Polly
E.Amazon Comprehend
AnswersD

Polly converts text to lifelike speech.

Why this answer

Amazon Comprehend can analyze sentiment or key phrases from reviews, and Amazon Polly converts text to speech. Transcribe does the opposite; Rekognition is for images; Translate is for translation.

Full explanation →
735
Multi-Selectmedium

A company is using Amazon Bedrock Agents to build a travel booking assistant that can search for flights, book hotels, and answer questions about travel policies. Which TWO components are required to enable the agent to call external services? (Select TWO.)

Select 2 answers
A.Action group with API schema
B.AWS Lambda function
C.Amazon DynamoDB table
D.Amazon SageMaker endpoint
E.Bedrock Knowledge Base
AnswersA, B

Action groups define the set of operations (APIs) the agent can call.

Why this answer

Action groups define the APIs (e.g., flight search, hotel booking) that the agent can invoke. A Lambda function provides the business logic to execute those API calls. Together they enable the agent to interact with external services.

Full explanation →
736
MCQeasy

What does the temperature parameter control in a text generation model?

A.The number of candidate tokens considered at each step
B.The degree of randomness in the generated output
C.The similarity to the training data distribution
D.The maximum number of tokens to generate
AnswerB

Temperature directly influences randomness: low values produce focused outputs, high values produce diverse outputs.

Why this answer

Temperature scales the logits before applying softmax; higher temperature produces more random outputs, lower temperature makes the model more deterministic.

Full explanation →
737
MCQeasy

A developer wants to quickly test different prompts and models for a text summarization task without writing any code. Which AWS service should they use?

A.Amazon Bedrock Playground
B.Amazon SageMaker Studio
C.Amazon Bedrock Agents
D.Amazon Bedrock Knowledge Bases
AnswerA

Bedrock Playground is an interactive console where you can test prompts and compare models without coding.

Why this answer

Amazon Bedrock Playground provides a no-code interface to experiment with models and prompts. Knowledge Bases and Agents are for building applications, not quick prototyping. SageMaker Studio is for full ML workflows.

Full explanation →
738
Multi-Selectmedium

A startup is building a semantic search system over their product catalog using Amazon Bedrock. They want to convert product descriptions into vector embeddings and store them in a vector database for similarity search. Which TWO actions should they take? (Select TWO.)

Select 2 answers
A.Store the embeddings in a vector database such as Amazon OpenSearch Serverless with the k-NN plugin
B.Use Amazon Titan Image Generator to create embeddings for each product image
C.Use Amazon Bedrock InvokeModel with Anthropic Claude to generate embeddings
D.Store the embedding vectors in an Amazon DynamoDB table as binary attributes
E.Use Amazon Titan Embeddings to generate vector embeddings from product descriptions
AnswersA, E

OpenSearch Serverless supports vector search with k-NN.

Why this answer

Amazon Titan Embeddings is designed for text embeddings. Amazon OpenSearch Serverless can serve as a vector database. The other options are not appropriate: image generation models, text generation models for embeddings, or Lambda for storage.

Full explanation →
739
Multi-Selecthard

Which TWO of the following are valid methods to reduce the risk of foundation models generating harmful or biased content?

Select 2 answers
A.Use a smaller model
B.Use a content filter
C.Apply prompt engineering to guide output
D.Fine-tune the model on a biased dataset
E.Disable all logging
AnswersB, C

Content filters can block harmful outputs.

Why this answer

Option B is correct because content filters act as a safety layer that intercepts and blocks harmful or biased outputs before they reach the user. These filters can be rule-based or use a separate classifier model trained to detect toxic, hateful, or biased language, reducing the risk of harmful content generation without altering the underlying model.

Exam trap

AWS often tests the misconception that simply using a smaller model or disabling logging can reduce bias, when in fact these actions either have no effect or worsen the problem, whereas content filters and prompt engineering are direct, effective mitigation strategies.

Full explanation →
740
MCQmedium

An e-commerce company uses Amazon Bedrock to generate product descriptions from keywords. Some descriptions contain inaccurate details about product specifications. Which approach should the company take to reduce factual errors?

A.Increase the maxTokens parameter to allow more detailed descriptions.
B.Use a different foundation model from Bedrock for each product category.
C.Deploy the model to a SageMaker endpoint and use human-in-the-loop validation.
D.Include the product specifications in the prompt and instruct the model to base the description on the provided data.
AnswerD

Providing facts in the prompt grounds the model's output and reduces fabrication.

Why this answer

Option D is correct because providing the product specifications directly in the prompt and instructing the model to base the description on that data grounds the generation in factual information, reducing hallucinations. This technique, known as prompt engineering with in-context learning, ensures the model uses the given data rather than relying on its training data, which may contain inaccuracies.

Exam trap

AWS often tests the misconception that increasing model parameters or changing models alone improves factual accuracy, when in fact prompt engineering with grounded data is the most effective and efficient method to reduce hallucinations.

How to eliminate wrong answers

Option A is wrong because increasing maxTokens only allows longer outputs but does not improve factual accuracy; it may even increase the chance of hallucinations by generating more unverified content. Option B is wrong because using a different foundation model for each category does not inherently reduce factual errors; all models can hallucinate, and this approach adds complexity without addressing the root cause of inaccurate specifications. Option C is wrong because deploying to a SageMaker endpoint with human-in-the-loop validation is an operational pattern for custom models, but it is overkill and inefficient for this use case; prompt engineering (Option D) is a simpler, more direct solution that avoids the latency and cost of human review for every generation.

Full explanation →
741
MCQeasy

A company wants to classify customer emails into categories (e.g., complaint, inquiry, feedback) using a foundation model. Which approach is MOST efficient?

A.Use Amazon Comprehend for custom classification
B.Train a custom model using Amazon SageMaker
C.Fine-tune a large language model on labeled emails
D.Use Amazon Lex with a classifier intent
AnswerA

Comprehend provides a ready-to-use classification API.

Why this answer

Amazon Comprehend provides a managed custom classification API that is purpose-built for text classification tasks like categorizing emails. It requires only a small set of labeled data to train a custom classifier, eliminating the need to manage infrastructure or fine-tune large models, making it the most efficient choice for this specific use case.

Exam trap

AWS often tests the misconception that any NLP task requires a large language model or custom training in SageMaker, when in fact managed services like Comprehend are optimized for common classification tasks and are more efficient.

How to eliminate wrong answers

Option B is wrong because training a custom model using Amazon SageMaker involves provisioning instances, managing training jobs, and handling model deployment, which is overkill and less efficient for a straightforward text classification task that can be handled by a managed service. Option C is wrong because fine-tuning a large language model (LLM) on labeled emails is computationally expensive, requires significant expertise in prompt engineering and hyperparameter tuning, and is not the most efficient approach when a simpler, purpose-built service like Comprehend exists. Option D is wrong because Amazon Lex is designed for building conversational chatbots and intent-based routing, not for batch or real-time text classification of emails; its classifier intent feature is meant for dialog management, not document categorization.

Full explanation →
742
MCQeasy

Which AWS service is BEST suited for extracting text from scanned PDF documents, such as invoices and receipts?

A.Amazon Comprehend
B.Amazon Textract
C.Amazon Rekognition
D.Amazon Transcribe
AnswerB

Textract is designed for OCR and data extraction from documents.

Why this answer

Amazon Textract is specifically designed to extract text, handwriting, and data from scanned documents. The other services serve different purposes: Rekognition for images/video, Comprehend for NLP, and Transcribe for audio-to-text.

Full explanation →
743
MCQmedium

A company wants to build a customer service chatbot that answers questions about their internal policy documents. The documents are updated monthly, and the team cannot afford to retrain a model each time. Which approach is MOST appropriate?

A.Train a custom model from scratch on the policy documents each month
B.Use Retrieval-Augmented Generation (RAG) with the policy documents indexed in a vector store
C.Fine-tune a base LLM on the policy documents monthly
D.Use a larger foundation model with a longer context window and paste all documents into each prompt
AnswerB

RAG retrieves relevant document chunks at query time, ensuring the chatbot always answers from the latest uploaded documents without any model retraining.

Why this answer

RAG (Retrieval-Augmented Generation) allows the LLM to retrieve relevant document sections at inference time, so knowledge stays current without retraining. The other options either require expensive retraining for each update or lack document grounding.

Full explanation →
744
MCQeasy

Refer to the exhibit. A security analyst is reviewing CloudTrail logs and notices a training job creation from an IP address (203.0.113.5) that is not associated with the company's network. What is the most likely cause?

A.The user john.doe is accessing the AWS Management Console from a VPN.
B.The CloudTrail log is being generated by a cross-account role.
C.The training job was created using the AWS CLI from an external machine.
D.The training job was created by a malicious actor who stole credentials.
AnswerA

A VPN would route traffic through an external IP; this is a common scenario for remote workers.

Why this answer

The IP address 203.0.113.5 is a non-routable test IP (RFC 5737) and not associated with the company's network. The most likely cause is that user john.doe is accessing the AWS Management Console through a VPN, which would route traffic through the VPN's public IP rather than the corporate network. This explains why the source IP appears external while the user identity is legitimate.

Exam trap

AWS often tests the distinction between 'external IP' and 'unauthorized access'—the trap here is assuming any external IP indicates a security breach, when in fact VPN usage is a legitimate and common cause for such logs.

How to eliminate wrong answers

Option B is wrong because cross-account roles would show the source IP of the role's session, not necessarily an external IP, and the log would include a 'userIdentity' with 'arn:aws:sts::...' indicating assumed role, which is not described. Option C is wrong because using the AWS CLI from an external machine would still show the machine's public IP, but the question states the IP is 'not associated with the company's network'—this is a plausible scenario but less likely than a VPN, as the user identity (john.doe) suggests a legitimate user, not an external machine. Option D is wrong because while stolen credentials are possible, the question asks for the 'most likely cause' given the context of a legitimate user identity; a malicious actor would typically not use a known corporate username without additional suspicious activity.

Full explanation →
745
MCQeasy

A developer wants to test different foundation models quickly without setting up infrastructure. Which AWS service allows interactive prompting and comparison of multiple models?

A.Amazon Comprehend
B.Amazon Bedrock Playground
C.Amazon Lex
D.Amazon SageMaker Studio
AnswerB

Bedrock offers a playground to interactively test and compare foundation models.

Why this answer

Amazon Bedrock provides a playground feature for testing models. Option A (SageMaker Studio) is for full notebook environment. Option C (Comprehend) is for analysis.

Option D (Lex) is for chatbots.

Full explanation →
746
MCQmedium

A data scientist is using Amazon SageMaker to train a model. The training job is taking longer than expected. Which change would most likely reduce training time?

A.Increase the number of training epochs
B.Use a larger batch size
C.Use a smaller instance type
D.Enable spot training
AnswerB

A larger batch size processes more samples per iteration, reducing the number of steps and overall time, provided the hardware supports it.

Why this answer

Using a larger batch size allows the model to process more training samples per iteration, which reduces the number of weight updates needed per epoch and can improve hardware utilization (e.g., GPU parallelism). This often leads to faster training times, provided the batch size fits within memory constraints and does not degrade model convergence.

Exam trap

The AIF-C01 exam often tests the misconception that reducing instance size or enabling spot instances directly improves training speed, when in fact these changes primarily affect cost or resource availability, not performance.

How to eliminate wrong answers

Option A is wrong because increasing the number of training epochs increases the total number of passes over the data, which would lengthen training time, not reduce it. Option C is wrong because using a smaller instance type reduces compute capacity (e.g., fewer vCPUs, less memory), which typically slows down training rather than speeding it up. Option D is wrong because enabling spot training (using Amazon EC2 Spot Instances) reduces cost but does not inherently reduce training time; it may even cause interruptions that delay completion.

Full explanation →
747
MCQeasy

A startup is building a customer support chatbot using Amazon Bedrock with the Claude foundation model. The chatbot needs to answer questions based on a knowledge base of frequently asked questions (FAQs) stored in an Amazon S3 bucket. The team wants to implement Retrieval Augmented Generation (RAG) to provide accurate and context-aware responses. They are evaluating different approaches to integrate the knowledge base. What is the most efficient way to implement RAG with Bedrock?

A.Use AWS Lambda to fetch documents from S3 and inject them into the prompt.
B.Manually extract all FAQs and include them in the prompt each time the chatbot responds.
C.Fine-tune the Claude model on the FAQs so the model memorizes the knowledge base.
D.Use Amazon Bedrock Knowledge Bases to directly connect the S3 bucket and retrieve relevant documents for the prompt.
AnswerD

Bedrock Knowledge Bases provides a managed RAG solution with automatic indexing and retrieval.

Why this answer

Option A is correct. Amazon Bedrock Knowledge Bases provides a native feature to connect to data sources like S3, automatically chunk and index documents, and retrieve relevant information. This is the most efficient and managed approach.

Option B is incorrect because manually including all FAQs in the prompt would exceed token limits and be impractical. Option C is incorrect because fine-tuning the model on FAQs is overkill for this use case and does not allow dynamic updates. Option D is a possible custom solution but is less efficient than using the built-in knowledge base feature.

Full explanation →
748
MCQeasy

A company needs to audit all API calls made to Amazon Bedrock, including model invocations and guardrail evaluations. Which AWS service should they enable to capture these API calls for compliance?

A.AWS CloudTrail
B.Amazon Macie
C.AWS Config
D.Amazon GuardDuty
AnswerA

CloudTrail logs all AWS API calls for auditing.

Why this answer

AWS CloudTrail records API calls for all AWS services, including Bedrock. It captures the caller identity, API, parameters, and response elements, which can be used for auditing and compliance.

Full explanation →
749
Multi-Selecthard

A company is deploying a machine learning model to detect fraudulent transactions. The dataset is highly imbalanced (1% fraud). The team needs to evaluate model performance and minimize false positives while maintaining high recall. Which TWO metrics should they focus on? (Select TWO.)

Select 2 answers
A.Accuracy
B.Recall
C.Precision
D.F1 score
E.AUC-ROC
AnswersB, C

Recall ensures high capture of fraud cases.

Why this answer

Precision measures how many of the predicted fraud cases are actually fraud (minimizing false positives). Recall measures how many actual fraud cases are caught. F1 balances both, but the question specifies minimizing false positives while maintaining high recall, so both precision and recall are directly relevant.

AUC-ROC and accuracy are less useful for imbalanced datasets.

Full explanation →
750
MCQmedium

A company is using Amazon Bedrock to generate images from text prompts. They need to ensure the generated images do not contain offensive content. Which feature should be enabled?

A.VPC endpoints
B.AWS WAF
C.Content moderation with AI
D.IAM policies
AnswerC

Bedrock's content moderation uses AI to detect and block offensive content.

Why this answer

Amazon Bedrock includes built-in content moderation that can filter harmful content in inputs and outputs. IAM policies (B) control access but not content. WAF (C) protects web applications.

VPC endpoints (D) secure network traffic.

Full explanation →
Page 9

Page 10 of 14

Page 11

All pages

1234567891011121314

Practice AIF-C01 by domain

Target a specific domain to shore up weak areas.

Applications of Foundation ModelsFundamentals of AI and MLFundamentals of Generative AIGuidelines for Responsible AISecurity, Compliance and Governance for AI Solutions
See all domains with question counts →
AWS Certified AI Practitioner AIF-C01 AIF-C01 Questions 676–750 | Page 10/14 | Courseiva