Free AWS Generative AI Developer - Professional Practice Questions

20 exam-style AWS Generative AI Developer - Professional questions with an explanation after every answer. No timer, no signup — about 15 minutes.

About the AWS Generative AI Developer - Professional exam

The AWS Generative AI Developer - Professional (AWS GenAI Dev) is a Professional-level certification. It covers foundation model integration, implementation and integration, AI safety and governance, operational efficiency, and testing and validation on AWS. The exam has 65 questions and runs 130 minutes. You need a scaled score of 700 out of 1000 to pass.

The 20 practice questions on this page are written by an AWS certified AI engineer and balanced across the exam domains using the official exam guide weighting. They use the same scenario-based style as the real AWS GenAI Dev exam.

Sample AWS Generative AI Developer questions

Three examples from the 20-question quiz. Answers stay hidden until you reveal them.

Sample Question 1AI Safety, Security, and Governance

An insurance company has deployed a policy information chatbot using Amazon Bedrock with a Knowledge Base containing all their policy documents. The chatbot must NEVER provide incorrect coverage amounts or policy terms, as errors could constitute regulatory violations. During testing, the model occasionally "fills in" details not present in the retrieved policy documents — for example, stating a deductible amount when the retrieved chunk only mentions the policy type without specific numbers. Which Bedrock Guardrails configuration best prevents this type of hallucination?

AConfigure a contextual grounding check with a high grounding threshold, which validates that each claim in the model's response is supported by the retrieved source content. Block responses that fall below the threshold.
BConfigure a denied topics policy that lists financial terms like "deductible," "premium," and "coverage amount" to prevent the model from including these terms unless they appear in the retrieved context
CConfigure content filters with the highest strictness level for all categories, and add a word filter that blocks responses containing specific dollar amounts
DConfigure an automated reasoning policy that uses formal logic to verify mathematical relationships between policy values mentioned in the response
Show answer
Answer: A — Configure a contextual grounding check with a high grounding threshold, which validates that each claim in the model's response is supported by the retrieved source content. Block responses that fall below the threshold.

Bedrock Guardrails' contextual grounding check is specifically designed to detect when a model's response makes claims not supported by the provided reference material (source documents). By setting a high grounding threshold, responses where the model fabricates details (like deductible amounts not in the source) are blocked. This is the native Bedrock solution for preventing RAG hallucination — it compares each statement in the response against the retrieved context and scores whether the response is grounded.

Sample Question 2Foundation Model Integration, Data Management, and Compliance

A developer is building a contract analysis application that extracts key terms, obligations, and deadlines from legal contracts using Amazon Bedrock. The initial implementation uses a zero-shot prompt that asks the model to "extract all important terms from this contract." Testing reveals that the model inconsistently identifies obligation types and misses nested deadline clauses. The developer needs to improve extraction accuracy while maintaining a manageable prompt that works across different contract formats. Which prompt engineering strategy should the developer implement?

ASwitch to a chain-of-thought prompt that instructs the model to first identify document sections, then analyze each section for obligations, and finally cross-reference deadlines, using XML tags to structure the output
BCreate a mega-prompt with 50 examples of correctly extracted contracts covering all possible contract formats, maximizing few-shot learning coverage
CUse Amazon Bedrock Prompt Management to create a versioned prompt template with 3-5 diverse few-shot examples, structured output schema using XML tags, and explicit extraction instructions for each term category including nested clauses
DFine-tune the model on 1,000 labeled contract extractions to teach it the specific extraction patterns, eliminating the need for complex prompting
Show answer
Answer: C — Use Amazon Bedrock Prompt Management to create a versioned prompt template with 3-5 diverse few-shot examples, structured output schema using XML tags, and explicit extraction instructions for each term category including nested clauses

Bedrock Prompt Management enables versioned, reusable prompt templates — ideal for iterating on extraction prompts. Using 3-5 diverse few-shot examples (rather than 50) provides sufficient pattern guidance without excessive token costs. Structured output with XML tags ensures consistent formatting across contract types. Explicit instructions for each term category (obligations, deadlines, nested clauses) address the specific failures observed. The versioning capability allows A/B testing of prompt improvements.

Sample Question 3Operational Efficiency and Optimization for GenAI Applications

A company has 15 different GenAI applications running on Amazon Bedrock across three AWS accounts. The CTO wants a centralized dashboard showing: (1) per-application token consumption and costs, (2) model latency percentiles (p50, p95, p99), (3) throttling rates and error rates per application, and (4) alerts when any application's daily spending exceeds its budget. Currently, no custom logging is implemented. Which monitoring architecture should the developer implement?

AEnable Amazon Bedrock model invocation logging in each account with logs sent to a centralized S3 bucket. Use Amazon Athena to query token usage from the logs, and create Amazon QuickSight dashboards for visualization. Set up CloudWatch Alarms on custom metrics published by a scheduled Lambda function that parses the logs.
BUse Amazon CloudWatch cross-account observability to aggregate Bedrock's built-in CloudWatch metrics (Invocations, InvocationLatency, InvocationThrottles) from all accounts into a central monitoring account. Create CloudWatch dashboards with per-application dimensions and CloudWatch Anomaly Detection for spending alerts.
CImplement application-level instrumentation using AWS X-Ray to trace all Bedrock API calls, sending traces to a centralized X-Ray group. Use X-Ray Analytics for latency analysis and configure X-Ray Insights for anomaly alerts.
DDeploy an Amazon Managed Grafana workspace connected to Amazon Managed Prometheus, with each application pushing custom metrics via the Prometheus remote write API for Bedrock usage data
Show answer
Answer: B — Use Amazon CloudWatch cross-account observability to aggregate Bedrock's built-in CloudWatch metrics (Invocations, InvocationLatency, InvocationThrottles) from all accounts into a central monitoring account. Create CloudWatch dashboards with per-application dimensions and CloudWatch Anomaly Detection for spending alerts.

Amazon Bedrock publishes built-in CloudWatch metrics including Invocations, InvocationLatency, InputTokenCount, OutputTokenCount, and InvocationThrottles with per-model dimensions. CloudWatch cross-account observability allows aggregating these metrics from all three accounts into a central monitoring account without custom logging. Per-application dimensions enable the dashboard requirements. CloudWatch Anomaly Detection can identify unusual spending patterns and trigger alerts. This leverages native AWS services with minimal custom code.

Want instant feedback and a weak-area summary?

Related AWS Generative AI Developer - Professional Practice Questions

Frequently asked questions about the AWS Generative AI Developer - Professional exam

What is the AWS Generative AI Developer certification?+

The AWS Generative AI Developer Professional certification validates advanced skills in building production-ready GenAI solutions using AWS services, especially Amazon Bedrock. It covers foundation model integration, prompt engineering at scale, RAG pipelines, AI safety, and operational efficiency.

How hard is the AWS Generative AI Developer exam?+

It is a Professional-level certification requiring deep hands-on experience with Amazon Bedrock and GenAI architecture patterns. It assumes familiarity with AWS services and production GenAI deployments. Most candidates study 4–6 weeks.

What topics are on the AWS GenAI Developer exam?+

The five domains are: FM Integration, Data Management, and Compliance (31%), Implementation and Integration (26%), AI Safety, Security, and Governance (20%), Operational Efficiency and Optimization (12%), and Testing, Validation, and Troubleshooting (11%).

Who should take the AWS Generative AI Developer certification?+

ML engineers, software engineers, and solutions architects who are building GenAI applications on AWS using Amazon Bedrock. It is best suited for those with hands-on Bedrock experience.

Want more practice?

Sign up free to access our full 500+ AWS Generative AI Developer - Professional question bank with quiz history tracking.