The AI Engineering Bible : Volume II: Building Production AI Systems
English | May 6, 2026 | ASIN: B0GZSMFCY9 | 622 pages | Epub | 428.99 KB
Build AI systems that hold up in production. Volume I taught you how to write prompts that work. Volume II teaches you how to ship them. Building Production AI Systems is the operational layer of the AI Engineering Bible - thirty-four chapters on system design, evaluation, retrieval, agents, observability, cost, and security. It's the book for engineers who have moved past "does it work in chat?" and now have to answer "will it work next Tuesday at 3 a.m. when traffic spikes?" What's inside System design. Orchestration patterns, request shapes, latency budgets, fallback strategies. Evals. What to measure, how to measure it, and how to keep evals honest as the model and prompt change. RAG. Chunking, embedding choice, reranking, hybrid search, freshness, and the multi-hop retrieval problem. Agents. Tool use, planning loops, ReAct, multi-agent coordination, human-in-the-loop gates. Observability. Tracing, prompt logging, eval pipelines, drift detection, cost telemetry. Cost engineering. Token budgets, model routing, prompt caching, batching, queuing. Security. Prompt injection defense, output validation, data exfiltration controls, red-team patterns. Vendor specifics. Where Claude, GPT, Gemini, and open-weight models actually differ in production. Companion to Volume I (foundations) and Volume III (the 2,160+ prompt library). Each volume stands alone; together they form the complete reference for serious AI engineers. Val Scancella is the author of AI Prompt Engineering: The 2026 Guide.
Links are Interchangeable - Single Extraction
