AI Security Engineering

AI Security Engineering is a practical, rigorous guide to building cybersecurity systems that use machine learning, large language models, AI agents, retrieval-augmented generation, MCP, and Python-without sacrificing security, reproducibility, or operational realism. Designed for students, researchers, engineers, and security professionals, the book moves from foundational concepts and intrusion detection to phishing, malware, IoT security, adversarial machine learning, explainability, calibration, federated learning, and production deployment. It also addresses the new risks created by LLM applications and autonomous agents, including prompt injection, tool misuse, memory poisoning, vector-store exposure, supply-chain threats, and AI incident response. Every major topic is connected to runnable code, realistic datasets, evaluation discipline, and deployment decisions. Readers learn not only how to build models, but how to test whether results are trustworthy, choose thresholds based on analyst capacity, detect leakage, manage drift, and create evidence that survives review. With case studies, deployment checklists, algorithm cards, dataset cards, exercises, and reproducible workflows, AI Security Engineering turns AI-powered security from an experiment into a defensible engineering practice for modern security teams operating in rapidly changing, adversarial, and increasingly agent-driven digital environments.