AWS Certified Generative AI Developer – Professional (AIP-C01) Study Guide: Domains and Study Plan
An AWS AIP-C01 guide covering the five exam domains, production generative AI integration, and a practical study plan.
The AWS Certified Generative AI Developer – Professional (AIP-C01) exam tests how to integrate foundation models and generative AI applications into production systems. Topics include RAG, agents, prompt management, data and compliance, security, cost optimization, evaluation, and troubleshooting. The focus is production engineering, not training foundation models.
Exam scope can change, so use the latest AWS AIP-C01 exam guide as your source of truth.
Who is this exam for? What background helps?
The exam is for engineers working on generative AI applications, platform integration, or governance. AWS describes its target candidate in terms of production application and generative AI experience; this is a profile, not a registration requirement. Before preparing, build the concepts covered by AIF-C01 and get comfortable with cloud, identity and security, application integration, and monitoring.
AIF-C01 focuses on AI concepts; MLA-C01 (check the exam version update) focuses on ML engineering; AIP-C01 focuses on production integration of generative AI systems.
Exam domains and weights
| Domain | Official name | Main focus | Weight |
|---|---|---|---|
| 1 | Foundation Model Integration, Data Management, and Compliance | Model integration, data management, and compliance | 31% |
| 2 | Implementation and Integration | Agents, models, and enterprise integration | 26% |
| 3 | AI Safety, Security, and Governance | AI safety, content safety, and governance | 20% |
| 4 | Operational Efficiency and Optimization for GenAI Applications | Operational efficiency, cost, and performance | 12% |
| 5 | Testing, Validation, and Troubleshooting | Testing, validation, and troubleshooting | 11% |
Domain 1 has the largest weight. Domain 5 has a smaller share, but production validation and troubleshooting still matter. Weights apply to scored content; use the official guide for detailed skills.
How to study each domain
Domain 1: Foundation Model Integration, Data Management, and Compliance
Start with business needs, model capabilities, and deployment constraints. Then address data sources, permissions, residency, and compliance. For vector databases and knowledge bases, consider data updates, indexing, and lifecycle—not only whether a query works.
Common mix-up: choosing a model before defining data and compliance requirements, or having no fallback when the model is unavailable. Document data sensitivity, allowed processing locations, quality requirements, and fallback options before selecting a model and retrieval design.
Domain 2: Implementation and Integration
Study agent and tool integration, model deployment strategies, FM APIs, enterprise systems, and application workflows. Understand the data flow among retrieval, models, tools, and business systems, including permissions and failure handling at each step.
Common mix-up: assuming a successful demo call means an integration is production-ready. Also plan for authentication, least privilege, idempotency, audit, timeouts, and human approval.
Domain 3: AI Safety, Security, and Governance
Learn how to protect model inputs and outputs, control harmful content and abuse, and apply identity, human oversight, and governance requirements in an application.
Common mix-up: configuring content guardrails while overlooking agent permissions, sensitive-data leakage, or human approval. Map each risk to a control, an owner, and audit evidence.
Domain 4: Operational Efficiency and Optimization
Compare model quality, latency, throughput, and cost. Understand how token use, caching, and model routing affect production systems, and use business metrics to check whether an optimization worked.
Common mix-up: optimizing only for cost or latency and degrading answer quality or business outcomes. Set quality, latency, and per-request cost targets before comparing alternatives.
Domain 5: Testing, Validation, and Troubleshooting
Build test sets, evaluation criteria, and regression checks. Diagnose whether an issue comes from the model, retrieval, prompts, tools, or downstream systems.
Common mix-up: judging output only by subjective impressions, or blaming the model for a retrieval failure. Use reproducible examples and a layered troubleshooting process.
Suggested study order
- Review the generative AI and foundation model concepts from AIF-C01.
- Study model, data, and compliance choices, then application and agent integration.
- Review security/governance and cost/performance trade-offs.
- Connect the topics through testing, evaluation, and troubleshooting scenarios.
Check your readiness with three questions
- How would you verify that RAG answers are grounded and run regression tests after a prompt or model update?
- How should identity, permissions, and human approval work when an agent calls an internal tool?
- When costs rise, how do you decide whether to change the model, context, caching, or call frequency?
If two answers are unclear, review those domains before adding more practice.
A three-step preparation plan
- Read the official guide. Mark familiar and weak areas against the tasks in all five domains.
- Map a production data flow. Identify sources, models, retrieval or tools, business actions, permissions, evaluation, and monitoring.
- Practice and review. Group mistakes by domain and cause, then revisit the matching official objectives.
Start practicing
When you are ready, use exam code AIP-C01 to practice by domain, then revisit the detailed skills in the official guide.