Production Generative AI on AWS is a decision-oriented field manual for the AWS Certified Generative AI Developer – Professional exam, paired with an AWS-verified Exam Simulator that scores you by domain. Written by a working practitioner, and kept current through the 2026 errata cycle.
Published Tuesday, the ninth of June, two thousand and twenty-six.
It’s live. Digital, PDF, Kindle, and paperback are all available now — on this site and on Amazon worldwide. The paperback prints on demand, so a copy ships wherever Amazon delivers.
Founding access: first 100 candidates get the Bundle for $39 (reg. $59) with code FOUNDING.
I spent more time editing this book than writing it. That is the wrong ratio for a blog post, and the correct ratio for a press.
The thick textbook and the short blog post are both well served. The thing in between, a tightly edited, decision-oriented reference for one platform, written by a practitioner, finishable on two flights, is not.
MineCloudCraft Press exists to fill that gap. Each title is between four hundred and seven hundred pages, single-author, scoped to one cloud or one platform, and edited toward a single working question: which option do I pick, and why? We are not vendor-neutral. We do not chase trends. We publish only when the platform is mature enough that the decisions inside the book will still be the right decisions twelve months after print.
The imprint is run by working practitioners, not career publishers. The editorial bar is simple: build for working engineers, refuse the marketing fog, write nothing the author has not personally shipped.
The complete field manual for the AWS Certified Generative AI Developer Professional exam. Bedrock, SageMaker, Knowledge Bases, Agents, Guardrails, and the decisions in between.
A practitioner’s reference for CloudWatch and the wider AWS observability stack. Metrics, logs, traces, alarms, dashboards, Synthetics, X-Ray, and the cost-and-cardinality decisions that quietly bankrupt teams.
Where the LLM stops being a novelty and starts being a tool on the toolbelt. AI-assisted code review, CI/CD pipelines that read their own logs, ChatOps with guardrails, root-cause analysis on real incidents, and the security boundaries that keep models out of the wrong systems.
Working titles are provisional. The two unscheduled books are being shaped with their authors through 2026; expect both to land before the end of 2027, or not at all. We would rather publish three good books a year than ten thin ones.
We are a small, careful imprint with a real distribution surface. Reach the right desk on the first try.