AWS Certified Generative AI Developer – Professional: A Field Reference and Study Guide for the Practitioner Sitting the Exam

For engineers who already ship production GenAI systems on AWS — and now need to demonstrate it on the exam. This is a field reference and study guide for the AWS Certified Generative AI Developer – Professional exam, written for senior engineers, architects, and the practitioners who have already built things on Bedrock and SageMaker. It is not an introduction. It assumes you have already invoked a foundation model, written a Lambda handler, configured IAM. The first chapter does not re-explain S3. It starts where the AWS documentation stops — at the architectural decisions, the operational trade-offs, the failure modes the docs do not name. What is in the book: Twenty-eight chapters in seven parts covering the full exam blueprint: foundations, designing GenAI solutions, building and customizing models, operations and observability, security and responsible AI, deployment and reliability, exam preparation. Six progressive hands-on labs that compose into one production-credible artifact — a multi-tenant retrieval-augmented generation system on Bedrock, fine-tuned with LoRA, hardened against prompt injection, deployed across regions with automated failover. Each lab is a complete walk-through with verification commands and teardown. A 75-question practice exam with detailed explanations for every answer. Questions distributed across the five domains in the same proportions as the real exam. Explanations name the reasoning pattern each question is testing, not just the right answer. Six reference appendices: an AWS service catalog scoped to GenAI workloads ("what it is / when to use / what it is not" for each service), an IAM policy cookbook with working JSON, a CloudWatch metric reference, a cost estimation worksheet, a glossary, and a curated further-reading list. The voice is the voice of someone who has built the systems, watched them fail, and remembers exactly why. Trade-offs are surfaced. Failure modes are named. Citations to original research are inline (Liu et al. on lost-in-the-middle, Greshake et al. on indirect prompt injection, Hu et al. on LoRA, Dettmers et al. on QLoRA, the SRE foundational texts). What is not here: the internals of transformer architecture, the mathematics of evaluation metrics, the introductory AWS material. Other books cover those, and the references appendix points to them. This book covers what someone with engineering experience needs to pass a hard exam while building knowledge that survives contact with production. If you have shipped systems and want a field reference that respects your time, this is the book. Read more