AWS Certified Generative AI Developer – Professional (AIP-C01) Study Guide [185 Minute Read]

A First-Principles Approach to Production GenAI on AWS

Welcome to a study guide built around understanding, not memorization. Every section answers three questions: why this topic matters in production, what mental model clarifies the mechanism, and what trap the exam uses to test whether you really understand it. By the end, you won't just know what Amazon Bedrock Knowledge Bases does — you'll know when to use it versus a custom OpenSearch implementation, and why that decision matters.

Official Exam Objectives: aws.amazon.com/certification/certified-generative-ai-developer-professional

Exam format: 75 questions (65 scored + 10 unscored) | 170 minutes | Passing score: 750/1000 | $300 USD
Question types: Multiple choice, multiple response, ordering, matching
Target candidate: 2+ years AWS experience, 1+ year hands-on GenAI development
Beta period: Active through March 31, 2026

Exam Domain Weights

Domain 1 dominates at 31% — but its concepts cascade through every other domain. The retrieval architecture you design in Domain 1 appears as a security consideration in Domain 3, a cost driver in Domain 4, and a quality question in Domain 5. Master Domain 1 and the other domains become easier to reason through.

---

(Table of Contents - For Reference)

• Phase 1: First Principles of Generative AI on AWS
  • 1.1. What Foundation Models Actually Are
    • 1.1.1. The FM as a Prediction Engine
    • 1.1.2. Context Windows and Token Mechanics
  • 1.2. How Generative AI Produces Output
    • 1.2.1. Inference Parameters and Output Control
    • 1.2.2. Embeddings and Semantic Representations
    • 1.2.3. Multimodal Inputs and Processing
  • 1.3. The Core Patterns: RAG and Agents
    • 1.3.1. Retrieval-Augmented Generation (RAG)
    • 1.3.2. AI Agents and Agentic Loops
  • 1.4. AWS GenAI Architecture Principles
    • 1.4.1. The Well-Architected GenAI Lens
    • 1.4.2. From PoC to Production
  • 1.5. Reflection Checkpoint
• Phase 2: Foundation Model Integration, Data Management & Compliance (31%)
  • 2.1. Selecting and Configuring Foundation Models
    • 2.1.1. Model Selection Criteria and Benchmarking
    • 2.1.2. The Bedrock Model Catalog: Claude, Titan, Llama, and Beyond
    • 2.1.3. FM Customization: Fine-Tuning and Continued Pre-Training
  • 2.2. FM Deployment and Lifecycle Management
    • 2.2.1. Deployment Strategies: On-Demand, Provisioned, and SageMaker Endpoints
    • 2.2.2. Dynamic Model Selection and Provider Switching
    • 2.2.3. Resilient FM Systems and Graceful Degradation
  • 2.3. Data Validation and Processing Pipelines
    • 2.3.1. Data Quality Validation for FM Consumption
    • 2.3.2. Multimodal Data Processing
    • 2.3.3. Input Formatting and Conversation Management
  • 2.4. Reflection Checkpoint
• Phase 3: Vector Stores, RAG, and Prompt Engineering (Domain 1 continued)
  • 3.1. Designing Vector Store Solutions
    • 3.1.1. Vector Database Options on AWS
    • 3.1.2. Amazon Bedrock Knowledge Bases Architecture
    • 3.1.3. High-Performance Vector Store Design
  • 3.2. Advanced Retrieval Mechanisms
    • 3.2.1. Chunking Strategies for Optimal Retrieval
    • 3.2.2. Embedding Model Selection and Management
    • 3.2.3. Search Architecture: Semantic, Keyword, and Hybrid
    • 3.2.4. Advanced Query Handling
  • 3.3. Prompt Engineering and Governance
    • 3.3.1. System Prompt Design and Instruction Frameworks
    • 3.3.2. Advanced Prompting Techniques
    • 3.3.3. Prompt Management and Versioning
    • 3.3.4. Prompt Flows and Chaining
  • 3.4. Reflection Checkpoint
• Phase 4: Implementation and Integration (26%)
  • 4.1. Agentic AI Solutions
    • 4.1.1. Amazon Bedrock Agents and Strands Agents
    • 4.1.2. Reasoning Patterns: ReAct and Chain-of-Thought
    • 4.1.3. Model Context Protocol (MCP)
    • 4.1.4. Human-in-the-Loop Workflows
  • 4.2. Enterprise Integration Architectures
    • 4.2.1. Event-Driven and Microservice Integration
    • 4.2.2. CI/CD Pipelines for GenAI Applications
    • 4.2.3. GenAI Gateway Architecture
  • 4.3. FM API Integration Patterns
    • 4.3.1. Synchronous and Asynchronous Invocation
    • 4.3.2. Streaming and Real-Time AI Systems
    • 4.3.3. Resilience: Retry, Backoff, and Fallback
    • 4.3.4. Intelligent Model Routing
  • 4.4. Reflection Checkpoint
• Phase 5: AI Safety, Security, and Governance (20%)
  • 5.1. Input and Output Safety Controls
    • 5.1.1. Amazon Bedrock Guardrails
    • 5.1.2. Hallucination Reduction and Output Validation
    • 5.1.3. Adversarial Input Defense
  • 5.2. Data Security and Privacy
    • 5.2.1. Network Isolation and Access Control
    • 5.2.2. PII Detection and Privacy Preservation
  • 5.3. AI Governance and Responsible AI
    • 5.3.1. Compliance Frameworks and Data Lineage
    • 5.3.2. Audit Logging and Source Tracking
    • 5.3.3. Responsible AI Principles in Production
  • 5.4. Reflection Checkpoint
• Phase 6: Operational Efficiency and Optimization (12%)
  • 6.1. Cost Optimization Strategies
    • 6.1.1. Prompt Caching and Token Optimization
    • 6.1.2. Semantic Caching and Response Reuse
    • 6.1.3. Provisioned Throughput and Batch Optimization
  • 6.2. Performance Optimization
    • 6.2.1. Retrieval Performance
    • 6.2.2. Generation Performance
    • 6.2.3. Cost-Performance Trade-off Analysis
  • 6.3. Monitoring and Observability
    • 6.3.1. CloudWatch Metrics and Alarms for GenAI
    • 6.3.2. Drift Detection and Continuous Evaluation
  • 6.4. Reflection Checkpoint
• Phase 7: Testing, Validation, and Troubleshooting (11%)
  • 7.1. Evaluation Frameworks and Quality Assessment
    • 7.1.1. RAG-Specific Evaluation: RAGAS Framework
    • 7.1.2. LLM-as-Judge and Human Evaluation
    • 7.1.3. A/B Testing and Incremental Rollout
  • 7.2. Troubleshooting GenAI Applications
    • 7.2.1. Diagnostic Methodology and Common Failure Patterns
    • 7.2.2. Performance Troubleshooting
    • 7.2.3. Quality Troubleshooting Runbook
  • 7.3. Reflection Checkpoint
• Phase 8: Exam Readiness
  • 8.1. Exam Strategy
  • 8.2. High-Frequency Exam Topics Quick Reference
  • 8.3. Practice Questions (Mixed Domain)
• Phase 9: Glossary
• Phase 10: Conclusion

---