Haystack 2 Pipelines: Modular RAG, Agents, and Evaluation in Practice

"Haystack 2 Pipelines: Modular RAG, Agents, and Evaluation in Practice"

For experienced Python developers, ML engineers, and AI platform builders, this book offers a rigorous guide to Haystack 2 as a production framework for modular LLM systems. Rather than treating RAG or agents as black-box recipes, it shows how to think in components, execution graphs, document stores, tools, and evaluators—so you can design systems that are understandable, testable, and adaptable under real operational pressure.

Across the book, readers will learn how to model documents, build indexing pipelines, enforce embedding compatibility, and engineer retrieval with sparse, dense, and hybrid strategies. It then moves through prompt construction, generator boundaries, end-to-end modular RAG composition, and the transition to agents, tool use, and stateful orchestration. A major strength of the book is its emphasis on evaluation: readers will leave with practical methods for statistical benchmarking, model-based judging, failure analysis, tracing, and continuous system improvement.

The treatment is deliberately advanced and implementation-oriented. It assumes familiarity with Python, LLM application patterns, and core retrieval concepts, and focuses on the design decisions, trade-offs, and failure modes that matter in practice. The result is a self-contained, architecture-first reference for building reliable Haystack 2 systems beyond demos.

© 2026 NobleTrex Press (Rafbók): 6610001244439

Útgáfudagur

Rafbók: 20 maj 2026