AWS Certified Generative AI Developer - Professional (AIP-C01) is a Professional-level certification that validates skills in designing, developing, and deploying generative AI applications on Amazon Bedrock. It spans the full breadth of GenAI application development -- RAG, agents, prompt engineering, and governance -- so theory alone rarely gets you through. Hands-on time with the Bedrock console and API matters a lot here.
Who Is AIP-C01 For?
This is a Professional-level certification for practitioners who actually design, build, and operate Bedrock-based GenAI applications in production.
| Target Audience | Why It Fits | |-----------------|-------------| | GenAI application developers | Validates hands-on skill building RAG and Agent apps with the Bedrock API/SDK | | ML engineers (post-MLA-C01) | Extends SageMaker-centric skills into the Bedrock GenAI space | | Solution architects | Builds the judgment needed for GenAI workload cost, security, and governance design | | Senior cloud engineers | Establishes credibility to lead GenAI adoption within a team |
The exam fee is $300 USD, and as a Professional-level exam, hands-on time in the Bedrock console and API matters more than theory alone.
---
Exam Basics
| Item | Detail | |------|--------| | Exam code | AIP-C01 | | Full name | AWS Certified Generative AI Developer - Professional | | Number of questions | 75 | | Duration | 180 minutes | | Passing score | 750 / 1000 | | Validity | 3 years | | Level | Professional |
75 questions in 180 minutes works out to about 2.4 minutes per question. Scenario-based questions dominate, so just reading each question takes real time -- flag anything you're unsure about and move on. Time management is essential.
The 5 Domains and Their Weight
!AIP-C01 Exam Domain Weights
Domain 1 - Foundation Model Integration, Data Management, and Compliance (31%)
The heaviest domain by far. It covers FM selection, comparison, and inference parameter tuning; the Amazon Bedrock Model Catalog; data preprocessing and ETL pipelines; vector databases, embeddings, and chunking strategies; RAG and Bedrock Knowledge Bases; and prompt engineering, templates, and caching -- the foundational skills for GenAI development.
Related posts: FM Model Selection and Solution Design, Data Pipelines and RAG Architecture, Prompt Engineering Practical Guide
Domain 2 - Implementation and Integration (26%)
Bedrock Agents and Action Groups, tool use and multi-agent orchestration, the Bedrock API / Converse API and streaming, Amazon Q Developer, model deployment and serverless inference, and microservices / event-driven integration are the core topics here.
Related posts: Bedrock Agents and Tool Integration, Model Deployment and API Integration
Domain 3 - AI Safety, Security, and Governance (20%)
Covers Bedrock Guardrails and content filtering, prompt injection defense and PII masking, KMS encryption / IAM / VPC isolation, bias detection, fairness, and explainability, SageMaker Clarify and Model Cards, and compliance frameworks.
Related posts: GenAI Safety and Security Controls, AI Governance and Responsible AI
Domain 4 - Operational Efficiency and Optimization for GenAI Applications (12%) / Domain 5 - Testing, Validation, and Troubleshooting (11%)
Combined, these two domains total 23% -- on par with the domains above. They cover Provisioned Throughput vs. on-demand, token-based cost optimization, CloudWatch / X-Ray monitoring, Bedrock Model Evaluation, ROUGE / BLEU / BERTScore metrics, and throttling / timeout debugging -- the full operations-and-evaluation surface.
Related post: Bedrock Cost Optimization and Model Evaluation
Study Strategy
If you've worked with Bedrock before
Make sure you understand the details you only pick up by touching the console and SDK directly -- the difference between InvokeModel and Converse API, the Knowledge Bases setup flow, how Guardrails policies are configured. Knowing the concept isn't the same as answering "how do you implement this in Bedrock."
If this is your first time
Start by clearly separating easily-confused concept pairs: the cost/latency trade-off between Provisioned Throughput and on-demand inference, and the difference in purpose between Guardrails (safety controls) and governance (compliance). Mixing these up can cost you nearly 20% on Domain 3 alone.
General strategy
Pair each service's limitations with the scenario it fits. Choose Bedrock Agents when you need autonomous multi-step execution, and Step Functions when you need explicit workflow orchestration. This "when to use what" judgment is the core of Domain 2 questions.
Exam Quick Reference
"Cost-sensitive but steady traffic" -- Provisioned Throughput "Unpredictable, bursty traffic" -- On-demand inference "Block prompt injection and harmful content" -- Bedrock Guardrails (input/output safety controls) "Bias detection, model transparency, compliance" -- AI Governance and Responsible AI "Multi-step task that autonomously calls external APIs" -- Bedrock Agents + Action Groups "Keep the grounding data behind generated answers current" -- RAG + Bedrock Knowledge Bases "Reduce cost for repeated identical requests" -- Prompt caching / response caching "Quantitatively evaluate model response quality" -- ROUGE, BLEU, BERTScore via Bedrock Model Evaluation
Next Steps
Once this guide gives you the big picture across all 5 domains, move on to the domain-specific deep dives to master each service's details, then test your readiness with the AIP-C01 practice exam.