AI Decision Engineering
An Independent Guide to Building Typed, Calibrated and Auditable AI Decisions into Software
-
- $0.99
-
- $0.99
Publisher Description
A practical, vendor-neutral and independent guide to using AI for decisions inside software rather than for free-form text. Much production AI chooses a category, scores a risk, ranks options, routes a request or answers yes or no, and the software acts on the answer. This book shows engineers how to make those decisions typed, calibrated, measurable and auditable. Following one fictional online marketplace, it covers framing decisions as typed questions, ground truth and decision metrics, rules and decision tables, classic classifiers, embeddings and nearest neighbors, language models with structured outputs, and specialized decision models. It explains probability calibration (Platt scaling, isotonic regression, temperature scaling, reliability diagrams and expected calibration error), cost-sensitive thresholds, abstention and human review, routing and triage cascades, and decisions inside agents. The final part covers drift and monitoring, decision logs and explanations, fairness and automated-decision law in several regions, and an architecture for latency, cost and vendor independence. Includes Python examples, checklists, a lab, study questions and a glossary.
This book is an independent publication and is not authorized, sponsored or endorsed by any company whose products it mentions. All trademarks belong to their respective owners and are used only for identification. This book was drafted with generative AI tools and checked with automated tools. Tools, features and prices change often; confirm them in current documentation.