Hands-on lab guide for the AWS Certified Generative AI Developer – Professional exam. 11 Jupyter notebooks covering all 4 exam domains with boto3, Amazon Bedrock, and real AWS infrastructure.
- Preparing for the AWS Certified Generative AI Developer – Professional (DEP-C01) exam
- Have an AWS account (pay-as-you-go, ~$25-38 total — see COST-GUIDE.md)
- Comfortable with Python and basic AWS concepts
- Clone this repo:
git clone https://github.com/btriani/aws-genai-lab-guide.git - Check prerequisites:
bash scripts/check-prerequisites.sh - Provision shared infrastructure:
python scripts/setup-resources.py - Open Lab 01:
labs/01-bedrock-foundation-models.ipynb
| # | Lab | Exam Domain (Weight) | Est. Cost | Est. Time |
|---|---|---|---|---|
| 01 | Bedrock Foundation Models | D1: Selection & Implementation (26%) | ~$0.35 | 45 min |
| 02 | Model Selection & Customization | D1 + D3: Selection + Optimization (26% + 24%) | ~$8-12 | 90 min |
| 03 | Prompt Engineering | D2: Building GenAI Apps (30%) | ~$0.30 | 60 min |
| 04 | Embeddings & Vector Search | D2: Building GenAI Apps (30%) | ~$2-3 | 60 min |
| 05 | RAG with Knowledge Bases | D2: Building GenAI Apps (30%) | ~$2-3 | 75 min |
| 06 | Bedrock Agents & Tool Use | D2: Building GenAI Apps (30%) | ~$2-3 | 75 min |
| 07 | Multi-Step GenAI Workflows | D2: Building GenAI Apps (30%) | ~$0.20 | 75 min |
| 08 | Model Evaluation | D3: Optimizing Performance (24%) | ~$2-3 | 60 min |
| 09 | Inference Optimization & Cost | D3: Optimizing Performance (24%) | ~$3-5 | 60 min |
| 10 | Guardrails & Responsible AI | D4: Security & Governance (20%) | ~$1-2 | 60 min |
| 11 | Security, Compliance & Logging | D4: Security & Governance (20%) | ~$1 | 60 min |
Total estimated cost: ~$25-38 | Total time: ~12 hours
See COST-GUIDE.md for detailed pricing breakdown.
| Domain | Weight | Labs |
|---|---|---|
| D1: Selection & Implementation of Foundation Models | 26% | 01, 02 |
| D2: Building Generative AI Applications | 30% | 03, 04, 05, 06, 07 |
| D3: Optimizing Performance & Inference | 24% | 08, 09 |
| D4: Security, Compliance & Governance | 20% | 10, 11 |
- notebook.ipynb — Jupyter notebook with interleaved markdown explanations and executable code cells. Each lab includes an overview, learning objectives, exam domain mapping, architecture diagram, lettered sections (A, B, C...), key takeaways, key concepts table, exam preparation Q&A, and cost breakdown.
Labs use AWS whitepapers (Well-Architected Framework, Generative AI on AWS, Bedrock User Guide, Shared Responsibility Model) as sample documents for RAG and embedding exercises. Documents are downloaded automatically by scripts/setup-resources.py. See assets/aws-whitepapers/README.md for details.
- Bedrock API Cheatsheet — InvokeModel, Converse, parameters, streaming, Knowledge Base and Agent APIs
- Exam Domains Cheatsheet — Domain breakdown, key concepts, study priority guide
- Services Comparison Cheatsheet — Decision trees: Bedrock vs SageMaker, vector store options, throughput modes
After running the labs, use the TEST-DRIVE.md to interactively test what you built: chat with models, ask your Knowledge Base questions about AWS whitepapers, trigger Guardrails with PII and blocked topics, and watch your Agent reason with tools. Each test includes example prompts and what results to expect — including abuse scenarios the exam covers.
For detailed instructions on where to find every resource in the AWS Console, see CONSOLE-GUIDE.md.
Every lab was tested end-to-end against real AWS infrastructure. We documented every issue we hit — deprecated model IDs, OpenSearch Serverless quirks, IAM gotchas, and more. See TROUBLESHOOTING.md before opening issues.
| Script | Purpose |
|---|---|
scripts/check-prerequisites.sh |
Verify AWS CLI, Python, credentials, and Bedrock access |
scripts/setup-resources.py |
Create S3 bucket, IAM roles, OpenSearch collection, download whitepapers |
scripts/cleanup-all.py |
Tear down all lab resources when done studying |
- DEP-C01 Exam Guide
- AWS Skill Builder — GenAI Learning Plan
- Amazon Bedrock Documentation
- AWS Workshops
Contributions welcome — open an issue or pull request.