HANDBOOK OF AI ENGINEERING AND AGENTIC AI: Building Intelligent AI Systems with LLMs, RAG, Multi-Agent Systems, and Production Workflows - Tapa blanda

Sinopsis

HANDBOOK OF AI ENGINEERING AND AGENTIC AI
Building an impressive AI prototype is easier than ever. Engineering an AI system that stays reliable, secure, measurable, and economical in production is a much harder problem.
Handbook of AI Engineering and Agentic AI is a rigorous, engineering-first guide to building intelligent systems that move beyond promising demos and perform dependably under real-world conditions. Prompts go brittle. Retrieval misses critical evidence. Structured outputs fail validation. Agents lose state, repeat actions, or exceed intended boundaries. This handbook develops the engineering principles and practical methods needed to address those failures systematically.
Inside, you will learn how to:

• Understand foundation-model behavior, context limits, decoding, caching, and cost trade-offs.
• Engineer prompts, structured outputs, and tool interfaces for dependable applications.
• Build retrieval-augmented systems using chunking, hybrid retrieval, reranking, and advanced retrieval architectures.
• Decide when retrieval, fine-tuning, or both are appropriate—and evaluate results using evidence rather than intuition.
• Design evaluation systems for quality, groundedness, robustness, safety, and cost.
• Build agentic systems that plan, use tools, maintain memory, recover from failures, and incorporate human oversight.
• Design multi-agent architectures in which specialized agents coordinate complex workflows.
• Apply security controls, guardrails, observability, and failure management to production systems.
• Optimize latency and cost while deploying and governing systems at scale.

"Sinopsis" puede pertenecer a otra edición de este libro.