AIF-C01 Practice Exam 58 — 20 Questions

Question 1

A developer is integrating an Amazon Bedrock foundation model into an application that must support multi-turn conversations, maintain chat history, and switch between different provider models with minimal code changes. The application should use a consistent request and response format. Which Amazon Bedrock API should the developer use?

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

Accepted Answer

The Converse API. The Converse API offers a unified conversational interface with a consistent message structure and response format across supported foundation models, simplifying multi-turn applications and reducing code changes when switching providers. InvokeModel requires provider-specific payloads, while model listing and customization job APIs serve discovery and training rather than conversational inference.

Answer

The InvokeModel API

Answer

The ListFoundationModels API

Answer

The CreateModelCustomizationJob API

Question 2

A logistics company deployed a demand forecasting model six months ago. The data science team notices that forecast accuracy has degraded gradually, and investigation shows that customer ordering patterns changed after a competitor entered the market. The team wants a repeatable process that detects when incoming data drifts from the training distribution and automatically retrains the model when drift exceeds a threshold. Which AWS approach should they implement?

Accepted Answer

Use Amazon SageMaker Model Monitor with a data quality baseline to detect drift, and trigger an AWS Lambda function from a CloudWatch alarm to start a SageMaker Pipelines retraining execution.. The requirement is automated drift detection on live traffic plus automatic retraining. SageMaker Model Monitor establishes a baseline from training data and continuously compares production inputs, raising CloudWatch violations when distributions diverge. Connecting those alarms to a Lambda function that starts a SageMaker Pipelines execution produces a repeatable detect-and-retrain loop, which is precisely the pattern the team needs after the market shift.

Answer

Use Amazon SageMaker Clarify to run a bias report on the training data weekly and retrain whenever the bias metric changes by more than five percent.

Answer

Configure AWS Glue DataBrew to profile the training dataset nightly and use AWS Step Functions to rebuild the feature store whenever a profile anomaly appears.

Answer

Enable Amazon CloudWatch Logs insights queries over the model endpoint logs and schedule a nightly EventBridge rule that restarts the endpoint when error rates rise.

Question 3

A financial services company uses Amazon Bedrock to power a customer-facing chatbot that answers questions about loan products. During testing, the team notices the model sometimes produces responses that sound confident but contain fabricated interest rates. The compliance team requires a mechanism to automatically detect when responses are not grounded in the company's approved product documentation. Which AWS capability should the team use to meet this requirement?

Accepted Answer

Amazon Bedrock Guardrails with contextual grounding checks. The requirement is to detect responses not supported by approved documentation, which is precisely what contextual grounding checks in Amazon Bedrock Guardrails do by scoring whether a response is grounded in the supplied reference text. Sentiment analysis, data drift monitoring, and API audit logging all operate on different layers and cannot judge whether a generated interest rate is factually supported.

Answer

Amazon SageMaker Model Monitor data drift detection

Answer

Amazon Comprehend sentiment analysis on the model output

Answer

AWS CloudTrail management event logging on the Bedrock endpoint

Question 4

A financial services company wants to deploy a generative AI assistant that summarizes internal earnings reports. The reports contain confidential, non-public financial data, and the security team requires that the data never leaves the company's AWS environment or be used to train the provider's public models. The company already uses Amazon Bedrock. Which characteristic of Amazon Bedrock directly satisfies this requirement?

Accepted Answer

Amazon Bedrock does not use customer inputs or outputs to train the base foundation models.. The security requirement is specifically that confidential data must not be used to train the provider's models. Amazon Bedrock's design keeps customer prompts and completions out of base foundation model training, which is the property that satisfies that constraint. Encryption, redaction, and regional pinning are useful but address different risks, so they do not meet the stated training-data isolation requirement.

Answer

Amazon Bedrock restricts all inference to a single AWS Region chosen at account creation.

Answer

Amazon Bedrock provides server-side encryption with AWS KMS keys for data at rest.

Answer

Amazon Bedrock automatically redacts personally identifiable information from all prompts before inference.

Question 5

A company is using Amazon Bedrock to build a conversational agent. They want to ensure the agent maintains context across multiple turns in a conversation. Which TWO strategies should the developer implement? (Choose two.)

Accepted Answer

Include the entire conversation history in each prompt sent to the model.. To maintain context across multiple turns, developers must include conversation history in each prompt. This can be done by sending the full history or by storing history externally and retrieving relevant parts. Foundation models are stateless, so context must be explicitly provided. Options suggesting built-in memory or session management are incorrect.

Answer

Use Amazon Bedrock's session management feature to automatically persist conversation state.

Answer

Enable the model's memory parameter to retain information across API calls.

Answer

Use a single API call with a streaming response to keep the connection open and maintain context.

Question 6

A media company wants to add a generative AI assistant to its internal knowledge portal. The assistant must answer employee questions using only the company's private policy documents, and it must cite the exact source passage for each answer. The team plans to use a foundation model hosted in Amazon Bedrock. Which approach should they implement to meet these requirements?

Accepted Answer

Use Retrieval Augmented Generation by storing policy documents in a vector store and retrieving relevant passages to include in the prompt sent to the model.. The requirements are grounding answers in private documents and citing exact source passages. Retrieval Augmented Generation satisfies both by indexing the policy corpus, retrieving the most relevant chunks at query time, and passing them into the prompt so the model can answer from supplied evidence. The application can then surface the retrieved passages as citations, keeping answers accurate and auditable without retraining the foundation model.

Answer

Raise the maximum token limit in the model invocation parameters so the entire policy library can be pasted into every prompt.

Answer

Fine-tune the foundation model on the full set of policy documents and deploy a custom model that answers from its updated weights.

Answer

Increase the model's temperature setting so it produces more varied and detailed answers from its pretrained knowledge.

Question 7

A developer sets a low temperature value when calling a foundation model through Amazon Bedrock for a use case that extracts structured fields from invoices. A colleague argues that lowering temperature reduces hallucination and guarantees correct extraction. How should the developer respond?

Accepted Answer

Lowering temperature makes sampling more deterministic, which improves consistency but does not guarantee factual correctness or eliminate hallucination.. Temperature governs the randomness of token sampling, so lowering it makes outputs more deterministic and repeatable. Determinism is not the same as correctness: a model can consistently produce the same wrong field value, and hallucinations can persist even at very low temperature. Accurate extraction therefore still depends on model capability, prompt design, and validation logic.

Answer

Lowering temperature forces the model to retrieve the correct field values from the invoice image before generating text.

Answer

Lowering temperature disables the model's ability to generate any text that was not present in its training data.

Answer

Lowering temperature guarantees the response will conform to the requested JSON schema without any prompt instructions.

Question 8

A machine learning engineer is preparing to fine-tune a foundation model for a specialized legal summarization task. The team has only a few thousand labeled examples and a limited budget. Which fine-tuning approach is most appropriate to adapt the model efficiently under these constraints?

Accepted Answer

Use parameter-efficient fine-tuning that updates only a small subset of adapter weights.. Parameter-efficient fine-tuning, such as LoRA, freezes the base model and trains only a small set of additional parameters. This cuts compute and memory requirements sharply, making adaptation feasible with limited data and budget. It still specializes the model for legal summarization, delivering the task adaptation the team needs without the expense of full retraining or training from scratch.

Answer

Increase the context window size so the model can read entire legal contracts at once.

Answer

Perform full fine-tuning that updates every parameter in the network.

Answer

Train the foundation model from scratch on the legal corpus.

Question 9

A junior data scientist is building a model to classify incoming customer support tickets into one of eight predefined categories such as Billing, Shipping, or Returns. Historical tickets already have correct category labels. Which type of machine learning is being used?

Accepted Answer

Supervised learning. The presence of correct historical labels plus a fixed set of eight output categories makes this multi-class classification, which belongs to supervised learning. The model is taught from known input-output pairs, then predicts the category for new tickets. Unsupervised, reinforcement, and generative approaches all lack the labeled input-output mapping central to this task.

Answer

Generative AI fine-tuning

Answer

Reinforcement learning

Answer

Unsupervised learning

Question 10

A content team wants to build an internal tool that drafts blog posts from short outlines. They have no machine learning engineers on staff and want AWS to manage the underlying model infrastructure while they focus only on prompts and output. Which AWS service should they use?

Accepted Answer

Amazon Bedrock. The team needs managed access to generative foundation models with minimal operational overhead, and Amazon Bedrock provides exactly that by serving multiple providers' models through one API without requiring the team to train or host anything. SageMaker AI offers more control but demands ML engineering, while Polly and Rekognition serve speech and vision tasks respectively.

Answer

Amazon SageMaker AI

Answer

Amazon Polly

Answer

Amazon Rekognition

Question 11

A developer is explaining how a large language model generates text so that a business stakeholder understands why the same prompt can yield different answers. Which description accurately captures the generation process?

Accepted Answer

The model predicts a probability distribution over the next token, samples from it, and appends the chosen token to repeat the process.. Text generation is autoregressive and probabilistic. At each step the model computes a distribution over the next token conditioned on everything before it, a token is sampled according to settings such as temperature and top-p, and the sequence grows one token at a time. Stochastic sampling is why identical prompts can yield different outputs, and why temperature changes affect variability.

Answer

The model queries an external search index and paraphrases the top-ranked web page for the prompt.

Answer

The model searches a stored database of previously written sentences and returns the closest exact match to the prompt.

Answer

The model compiles the prompt into a set of if-then rules that deterministically map keywords to canned responses.

Question 12

A company is using Amazon Bedrock to power a customer-facing chatbot. The security team wants to ensure that the chatbot does not respond to prompts that attempt to elicit inappropriate or off-topic responses, such as discussing competitors or providing medical advice. The company also wants to monitor and log all blocked prompts for analysis. Which AWS service or feature should they use to achieve this?

Accepted Answer

Amazon Bedrock Guardrails with denied topics and content filters. Amazon Bedrock Guardrails provides configurable safeguards, including denied topics and content filters, that can block inappropriate or off-topic responses in real time. It also supports logging of blocked prompts, which allows the security team to monitor and analyze attempts. This is the most direct and effective solution for the chatbot's requirements.

Answer

AWS WAF with custom rules to block inappropriate prompts

Answer

Amazon SageMaker Ground Truth to label and filter prompts

Answer

Amazon Comprehend to analyze and filter prompts

Question 13

A company is using Amazon SageMaker to deploy a machine learning model for credit scoring. The compliance team requires that the model's predictions be explainable to customers, and that the company can demonstrate which features contributed most to a decision. Which SageMaker feature should be used to meet this requirement?

Accepted Answer

SageMaker Clarify. The requirement is to explain individual predictions by identifying which features contributed most to a credit decision. SageMaker Clarify offers SHAP-based feature attributions that can be generated for real-time predictions. This enables the company to provide explanations to customers and demonstrate compliance. Other SageMaker features focus on monitoring, debugging, or experiment tracking, not on model explainability.

Answer

SageMaker Debugger

Answer

SageMaker Experiments

Answer

SageMaker Model Monitor

Question 14

A retailer wants to group its customers into distinct behavioral segments for targeted marketing, but it has no predefined segment labels and no historical outcomes to learn from. Which machine learning approach should the retailer use?

Accepted Answer

Clustering. Because no segment labels exist and the objective is to discover natural groupings among customers, this is an unsupervised learning problem best solved with clustering. Classification, regression, and reinforcement learning all depend on labels, targets, or reward signals that the retailer does not have, so clustering is the only approach that fits the described situation.

Answer

Reinforcement learning

Answer

Regression

Answer

Supervised classification

Question 15

A support organization notices its generative AI assistant sometimes invents policy details that do not exist. Leadership wants a measurable way to track whether this behavior improves over time. Which action should the team take?

Accepted Answer

Build a labeled evaluation set of questions with verified answers and score responses for factual consistency on a recurring basis.. Turning an observed quality problem into a tracked metric requires a fixed evaluation set with verified answers and a scoring procedure that can be rerun after each change. This makes comparisons across prompts, retrieval configurations, and models meaningful, and it establishes a regression baseline so the organization can prove whether fabrication is actually declining.

Answer

Switch to a model with a larger parameter count and assume the fabrication problem disappears.

Answer

Increase the model's maximum output token limit so it has more room to explain itself.

Answer

Rely on customer complaints as the primary signal for whether fabricated policy details are decreasing.

Question 16

A financial services company wants to deploy a generative AI assistant that answers questions using its internal policy documents. The team is concerned that the model may invent policies that do not exist. Which approach best reduces this risk while keeping the model's answers tied to the company's actual documents?

Accepted Answer

Use retrieval augmented generation to fetch relevant document passages and include them in the prompt.. Retrieval augmented generation grounds the model by supplying relevant passages from the company's own document repository at inference time. Because the model conditions its answer on retrieved, verifiable content, it is far less likely to invent policies. This directly satisfies the requirement that answers stay tied to actual internal documents rather than relying on the model's parametric memory.

Answer

Switch to a larger foundation model with more parameters to improve factual recall.

Answer

Lower the maximum token limit so the model produces shorter answers with fewer chances to err.

Answer

Increase the model's temperature setting so it explores more diverse phrasings.

Question 17

A startup wants to add image generation to its design tool. The team does not want to manage GPU infrastructure, train models, or host endpoints, and they want to call a managed API for a text-to-image foundation model available in Amazon Bedrock. Which action should they take?

Accepted Answer

Use an Amazon Bedrock image generation model through the InvokeModel API with a text prompt.. Amazon Bedrock exposes image generation foundation models such as Amazon Titan Image Generator as managed, serverless APIs invoked with a text prompt. This delivers text-to-image capability without GPU provisioning, training, or endpoint management. SageMaker self-hosting adds operational work, while Rekognition analyzes images and Polly synthesizes speech, so neither can generate design images from text.

Answer

Use Amazon Polly to convert text prompts into images for the design tool.

Answer

Use Amazon Rekognition to synthesize new design images from text descriptions.

Answer

Deploy an open-source diffusion model on Amazon SageMaker endpoints and manage scaling themselves.

Question 18

A logistics company is planning its first machine learning project and the leadership team asks which statements correctly describe fundamental machine learning concepts. Which TWO statements are accurate? (Choose two.)

Accepted Answer

In supervised learning, the model learns from input-output pairs where the desired output is provided during training.. Supervised learning is defined by training on input-output pairs with known targets, while unsupervised learning finds structure in data without any labels. Both statements describe core, vendor-neutral machine learning concepts that a logistics team should understand before scoping its first project.

Answer

A machine learning model's predictions are deterministic rules written by engineers rather than patterns inferred from data.

Answer

Reinforcement learning requires a fully labeled dataset of correct actions for every possible situation before training can begin.

Answer

Model training always improves accuracy on unseen data as more epochs are run, so training should continue until loss reaches zero.

Question 19

A product team wants its generative AI assistant to answer questions about internal policy documents accurately rather than from the model's general training knowledge. Which TWO techniques should they apply? (Choose two.)

Accepted Answer

Instruct the model to answer only from the provided context and to state when the context is insufficient.. Grounding combines retrieval of authoritative source passages with an instruction that limits the model to that evidence and requires an explicit fallback when evidence is missing. Together these reduce reliance on parametric knowledge and make answers traceable to internal documents, which is what accurate policy question answering demands.

Answer

Increase the temperature setting so the model explores a wider range of possible answers.

Answer

Shorten the system prompt to a single sentence to reduce token consumption.

Answer

Ask the model to produce the longest possible answer for every question.

Question 20

A product team wants to compare two foundation models on their own customer-support transcripts before choosing one for a chatbot. They need a quantitative measure of how well each model's answers match reference answers. Which evaluation approach fits this need?

Accepted Answer

Automated evaluation using similarity metrics such as ROUGE or BERTScore between model outputs and reference answers. Automated text-similarity metrics score generated answers against reference answers and yield numbers that can be compared across models on the same transcripts. Human ratings are qualitative and costly, latency and cost measure operations rather than quality, and public benchmarks do not reflect the company's domain data. Automated evaluation on in-domain data is the fit.

Answer

Human evaluation where reviewers rate each answer on a five-point scale

Answer

Reviewing each model's published model card and benchmark leaderboard rankings

Answer

Measuring inference latency and cost per thousand tokens for each model