AWS Certified Generative AI Developer - Professional Practice Exam

Exam Details

• Exam Code: AIP-C01
• Category: Professional
• Exam Duration: 180 minutes
• Exam Format: 75 questions
• Question Types: Multiple choice or multiple response
• Passing Score: 750 (on a scale of 100–1,000)
• Exam Languages: English, Japanese, Korean, Simplified Chinese

Course Outline

The AWS Certified Generative AI Developer – Professional (AIP-C01) exam covers the following topics - 

Domain 1: Foundation Model Integration, Data Management, and Compliance (31%) 

Task 1.1 — Analyse Requirements and Design GenAI Solutions

• Create comprehensive architectural designs aligned with specific business needs and technical constraints, encompassing appropriate FM selection, integration patterns, and deployment strategies
• Develop technical proof-of-concept implementations to validate feasibility, performance characteristics, and business value prior to full-scale deployment (e.g., using Amazon Bedrock)
• Create standardised technical components to ensure consistent implementation across multiple deployment scenarios using the AWS Well-Architected Framework and the Generative AI Lens

Task 1.2 — Select and Configure Foundation Models

• Assess and select foundation models for optimal alignment with specific business and technical requirements using performance benchmarks, capability analysis, and limitation evaluation
• Create flexible architecture patterns to enable dynamic model selection and provider switching without requiring code modifications (e.g., using AWS Lambda, Amazon API Gateway, AWS AppConfig)
• Design resilient AI systems for continuous operation during service disruptions using AWS Step Functions circuit breaker patterns and Amazon Bedrock Cross-Region Inference
• Implement FM customisation and lifecycle management through fine-tuning techniques such as LoRA, model versioning via SageMaker Model Registry, automated deployment pipelines, and rollback strategies for failed deployments

Task 1.3 — Implement Data Validation and Processing Pipelines

• Build comprehensive data validation workflows to ensure data meets quality standards for FM consumption using AWS Glue Data Quality, SageMaker Data Wrangler, and Amazon CloudWatch metrics
• Create data processing pipelines for complex data types including text, images, audio, and tabular data using Amazon Bedrock multimodal models, SageMaker Processing, and AWS Transcribe
• Format input data for FM inference in accordance with model-specific requirements — including JSON formatting for Amazon Bedrock API requests and conversation formatting for dialogue-based applications
• Enhance input data quality to improve FM response consistency using Amazon Comprehend for entity extraction and Lambda functions for data normalisation

Task 1.4 — Design and Implement Vector Store Solutions

• Create advanced vector database architectures for efficient semantic retrieval using Amazon Bedrock Knowledge Bases, Amazon OpenSearch Service with the Neural plugin, Amazon RDS, and Amazon DynamoDB
• Develop comprehensive metadata frameworks to improve search precision and context awareness for FM interactions using S3 object metadata, custom attributes, and domain classification tagging
• Implement high-performance vector indexing strategies including OpenSearch sharding, multi-index approaches, and hierarchical indexing for optimised semantic search at scale
• Design automated data maintenance systems using incremental update mechanisms, real-time change detection, and scheduled refresh pipelines to ensure vector stores remain current and accurate

Task 1.5 — Design Retrieval Mechanisms for FM Augmentation

• Develop effective document segmentation strategies using Amazon Bedrock chunking capabilities, fixed-size chunking via Lambda functions, and hierarchical chunking based on content structure
• Select and configure optimal embedding solutions such as Amazon Titan based on dimensionality, domain fit, and performance characteristics of Amazon Bedrock embedding models
• Deploy vector search solutions using Amazon OpenSearch Service, Amazon Aurora with the pgvector extension, and Amazon Bedrock Knowledge Bases with managed vector store functionality
• Create advanced hybrid search architectures combining keyword and semantic search using OpenSearch and Amazon Bedrock reranker models for improved retrieval relevance and accuracy
• Design sophisticated query handling systems for improved retrieval effectiveness using Amazon Bedrock for query expansion, Lambda functions for query decomposition, and Step Functions for query transformation
• Develop standardised access mechanisms for seamless FM integration using Model Context Protocol (MCP) clients and function calling interfaces for vector queries

Task 1.6 — Implement Prompt Engineering Strategies and Governance

• Create effective model instruction frameworks using Amazon Bedrock Prompt Management to enforce role definitions and Amazon Bedrock Guardrails to enforce responsible AI guidelines
• Build interactive AI systems to maintain context and improve user interactions using Step Functions for clarification workflows, Amazon Comprehend for intent recognition, and DynamoDB for conversation history storage
• Implement comprehensive prompt governance systems including parameterised templates, approval workflows, S3 template repositories, CloudTrail usage tracking, and CloudWatch Logs access logging
• Develop quality assurance systems for prompt validation using Lambda functions, Step Functions for edge case testing, and CloudWatch for prompt regression testing
• Apply advanced iterative prompt refinement using structured input components, chain-of-thought instruction patterns, output format specifications, and feedback loops
• Design complex prompt chains using Amazon Bedrock Prompt Flows with conditional branching, reusable components, and integrated pre- and post-processing steps

Domain 2: Implementation and Integration (26%) 

Task 2.1 — Implement Agentic AI Solutions and Tool Integrations

• Develop intelligent autonomous systems with appropriate memory and state management capabilities using Strands Agents, AWS Agent Squad, and Model Context Protocol (MCP) for agent-tool interactions
• Create advanced problem-solving systems using AWS Step Functions to implement ReAct patterns and chain-of-thought reasoning approaches
• Develop safeguarded AI workflows with controlled FM behaviour using Step Functions stopping conditions, Lambda timeout mechanisms, IAM policy enforcement, and circuit breakers
• Design sophisticated multi-model coordination systems using specialised FMs for complex tasks, custom aggregation logic, and model selection frameworks
• Develop human-in-the-loop review processes using Step Functions to orchestrate approval workflows and API Gateway to implement feedback collection mechanisms
• Implement intelligent custom tool integrations using the Strands API with standardised function definitions and Lambda functions for error handling and parameter validation
• Build model extension frameworks using Lambda for lightweight stateless MCP servers and Amazon ECS for complex tool-serving MCP servers

Task 2.2 — Implement Model Deployment Strategies

• Deploy foundation models based on specific application needs using Lambda for on-demand invocation, Amazon Bedrock provisioned throughput configurations, and SageMaker AI endpoints for hybrid solutions
• Address the unique deployment challenges of large language models including container-based deployment patterns optimised for memory requirements, GPU utilisation, and token processing capacity
• Develop optimised FM deployment approaches using smaller pre-trained models for specific tasks and API-based model cascading to efficiently handle routine queries

Task 2.3 — Design and Implement Enterprise Integration Architectures

• Create enterprise connectivity solutions using API-based integrations with legacy systems, event-driven architectures for loose coupling, and data synchronisation patterns
• Develop integrated AI capabilities for existing applications using API Gateway for microservice integrations, Lambda functions for webhook handlers, and Amazon EventBridge for event-driven integrations
• Create secure access frameworks using identity federation between FM services and enterprise systems, role-based access controls, and least-privilege API access policies
• Deploy cross-environment AI solutions using AWS Outposts for on-premises data integration, AWS Wavelength for edge deployments, and secure routing between cloud and on-premises resources
• Implement CI/CD pipelines with centralised GenAI gateway architectures using AWS CodePipeline and AWS CodeBuild, incorporating security scans, automated testing frameworks, and rollback support

Task 2.4 — Implement FM API Integrations

• Build flexible model interaction systems using Amazon Bedrock APIs for synchronous and asynchronous processing via language-specific AWS SDKs and Amazon SQS
• Develop real-time AI interaction systems using Amazon Bedrock streaming APIs, WebSockets, server-sent events, and API Gateway chunked transfer encoding for immediate FM feedback
• Create resilient FM systems with exponential backoff using the AWS SDK, API Gateway rate limiting, fallback mechanisms for graceful degradation, and AWS X-Ray for cross-service observability
• Develop intelligent model routing systems using Step Functions for dynamic content-based routing, API Gateway with request transformations, and metrics-based intelligent routing to specialised FMs

Task 2.5 — Implement Application Integration Patterns and Development Tools

• Create FM API interfaces capable of handling streaming responses, token limit management, and retry strategies for model timeouts using Amazon API Gateway
• Develop accessible AI interfaces using AWS Amplify for declarative UI components, OpenAPI specifications for API-first development, and Amazon Bedrock Prompt Flows for no-code workflow builders
• Create business system enhancements using Lambda for CRM integrations, Step Functions for document processing orchestration, Amazon Q Business for internal knowledge tools, and Amazon Bedrock Data Automation for automated data processing workflows
• Enhance developer productivity using Amazon Q Developer for code generation, refactoring, API assistance, and AI component testing
• Develop advanced GenAI applications using Strands Agents and AWS Agent Squad for native orchestration, Step Functions for agent design patterns, and Amazon Bedrock for prompt chaining
• Improve troubleshooting efficiency using CloudWatch Logs Insights for prompt and response analysis and AWS X-Ray for FM API call tracing

Domain 3: AI Safety, Security, and Governance (20%)

Task 3.1 — Implement Input and Output Safety Controls

• Develop comprehensive content safety systems using Amazon Bedrock Guardrails to filter harmful user inputs, with custom moderation workflows via Step Functions and Lambda functions
• Create content safety frameworks to prevent harmful FM outputs using Amazon Bedrock Guardrails, specialised FM evaluations for toxicity detection, and text-to-SQL transformations for deterministic results
• Develop accuracy verification systems to reduce hallucinations using Amazon Bedrock Knowledge Base for response grounding, confidence scoring, semantic similarity search, and JSON Schema for structured outputs
• Create multi-layered defence-in-depth safety systems using Amazon Comprehend for pre-processing filters, model-based guardrails, Lambda for post-processing validation, and API Gateway for response filtering
• Implement advanced threat detection against prompt injection, jailbreak attempts, and adversarial inputs using input sanitisation, safety classifiers, and automated adversarial testing workflows

Task 3.2 — Implement Data Security and Privacy Controls

• Design protected AI environments using VPC endpoints for network isolation, IAM policies for secure data access, AWS Lake Formation for granular access control, and CloudWatch for access monitoring
• Develop privacy-preserving systems using Amazon Comprehend and Amazon Macie for PII detection, Amazon Bedrock native data privacy features, Guardrails for output filtering, and S3 Lifecycle configurations for data retention
• Create privacy-focused AI systems using data masking techniques, anonymisation strategies, and Amazon Bedrock Guardrails to protect sensitive user information while maintaining FM utility

Task 3.3 — Implement AI Governance and Compliance Mechanisms

• Develop compliance frameworks using SageMaker model cards, AWS Glue for automated data lineage tracking, metadata tagging for data source attribution, and CloudWatch Logs for comprehensive decision logging
• Implement data source traceability using AWS Glue Data Catalog for source registration, metadata tagging for FM-generated content attribution, and CloudTrail for audit logging
• Create organisational governance systems aligned with organisational policies, regulatory requirements, and responsible AI principles
• Implement continuous monitoring and advanced governance controls including bias drift detection, automated alerting and remediation workflows, token-level redaction, and AI output policy filters

Task 3.4 — Implement Responsible AI Principles

• Develop transparent AI systems using reasoning displays for user-facing explanations, CloudWatch for confidence metrics, evidence presentation for source attribution, and Amazon Bedrock agent tracing for reasoning traces
• Apply fairness evaluations using pre-defined CloudWatch fairness metrics, Amazon Bedrock Prompt Management for systematic A/B testing, and LLM-as-a-judge automated model evaluations
• Build policy-compliant AI systems using Amazon Bedrock Guardrails based on policy requirements, model cards to document FM limitations, and Lambda functions for automated compliance checks

Domain 4: Operational Efficiency and Optimisation for GenAI Applications (12%)

Task 4.1 — Implement Cost Optimisation and Resource Efficiency Strategies

• Develop token efficiency systems using context window optimisation, prompt compression, context pruning, and response size controls to reduce FM costs while maintaining effectiveness
• Create cost-effective model selection frameworks evaluating cost-capability tradeoffs, tiered FM usage by query complexity, and price-to-performance ratio measurement
• Develop high-performance FM systems using batching strategies, capacity planning, utilisation monitoring, auto-scaling configurations, and provisioned throughput optimisation
• Create intelligent caching systems using semantic caching, result fingerprinting, edge caching, deterministic request hashing, and prompt caching to minimise unnecessary FM invocations

Task 4.2 — Optimise Application Performance

• Create responsive AI systems using pre-computation for predictable queries, latency-optimised Amazon Bedrock models for time-sensitive applications, parallel requests for complex workflows, and response streaming
• Enhance retrieval performance through index optimisation, query preprocessing, and hybrid search implementation with custom relevance scoring
• Implement FM throughput optimisation using token processing strategies, batch inference, and concurrent model invocation management
• Enhance FM output quality using model-specific parameter configurations, A/B testing for improvement evaluation, and appropriate temperature and top-k/top-p sampling parameter selection
• Create efficient resource allocation systems using capacity planning for token processing, utilisation monitoring, and auto-scaling configurations optimised for GenAI traffic patterns
• Optimise FM system performance using API call profiling, vector database query optimisation, latency reduction techniques specific to LLM inference, and efficient service communication patterns

Task 4.3 — Implement Monitoring Systems for GenAI Applications

• Create holistic observability systems encompassing operational metrics, performance tracing, FM interaction tracing, and business impact metrics via custom dashboards
• Implement comprehensive GenAI monitoring using CloudWatch to track token usage, prompt effectiveness, hallucination rates, response quality, and cost anomalies; and Amazon Bedrock Model Invocation Logs for detailed request and response analysis
• Develop integrated observability solutions providing compliance monitoring, forensic traceability, audit logging, user interaction tracking, and model behaviour pattern analysis
• Create tool performance frameworks using call pattern tracking, performance metric collection, tool calling observability, and usage baselines for anomaly detection
• Build vector store operational management systems using performance monitoring, automated index optimisation routines, and data quality validation processes
• Develop FM-specific troubleshooting frameworks using golden datasets for hallucination detection, output diffing for response consistency analysis, and reasoning path tracing for logical error identification

Domain 5: Testing, Validation, and Troubleshooting (11%)

Task 5.1 — Implement Evaluation Systems for GenAI

• Develop comprehensive assessment frameworks evaluating FM outputs across metrics including relevance, factual accuracy, consistency, and fluency — going beyond traditional ML evaluation approaches
• Create systematic model evaluation systems using Amazon Bedrock Model Evaluations, A/B and canary testing of FMs, multi-model evaluation, and cost-performance analysis including token efficiency and latency-to-quality ratios
• Develop user-centred evaluation mechanisms using feedback interfaces, rating systems for model outputs, and annotation workflows to continuously assess response quality
• Implement continuous quality assurance processes with regression testing for model outputs and automated quality gates for deployments
• Create comprehensive multi-perspective assessment systems using RAG evaluation, LLM-as-a-judge automated quality assessment, and human feedback collection interfaces
• Implement retrieval quality testing using relevance scoring, context matching verification, and retrieval latency measurements to evaluate and optimise information retrieval components
• Develop agent performance frameworks using task completion rate measurements, tool usage effectiveness evaluations, Amazon Bedrock Agent evaluations, and reasoning quality assessment in multi-step workflows
• Create comprehensive reporting systems using visualisation tools, automated reporting mechanisms, and model comparison visualisations to communicate performance metrics to stakeholders
• Build deployment validation systems using synthetic user workflows, AI-specific output validation for hallucination rates, semantic drift detection, and automated quality checks for response consistency

Task 5.2 — Troubleshoot GenAI Applications

• Resolve content handling issues such as context window overflow and truncation-related errors using dynamic chunking strategies, prompt design optimisation, and overflow diagnostic techniques
• Diagnose and resolve FM integration issues specific to GenAI services using error logging, request validation, and response analysis
• Troubleshoot prompt engineering problems using prompt testing frameworks, version comparison methodologies, and systematic prompt refinement approaches
• Diagnose retrieval system failures including embedding quality issues, vector search performance degradation, drift monitoring, vectorisation problems, chunking remediation, and search optimisation
• Resolve prompt maintenance issues using CloudWatch Logs to diagnose prompt confusion, AWS X-Ray for observability pipelines, schema validation to detect format inconsistencies, and systematic prompt refinement workflows