knowledge-centric-ai

RAG Field Guide

A Practical Reference for Retrieval-Augmented Generation Systems

Type

Field Guide

Version

1.0 · July 2026

Author

Mariano Garralda-Barrio

Overview

The RAG Field Guide is a practical handbook for engineers, researchers and architects designing Retrieval-Augmented Generation (RAG) systems.

Rather than presenting RAG as a collection of isolated techniques, the guide organizes the design space around the engineering decisions required to build reliable, production-ready systems.

It provides a concise reference covering retrieval strategies, ranking methods, architecture families, evaluation methodologies, production considerations and engineering trade-offs from an evidence-centric perspective.

The guide complements the Knowledge-Centric AI research papers and white papers by providing a practical engineering reference for the design, evaluation and operation of modern RAG systems.

The handbook includes:

RAG architecture families
Retrieval methods
Ranking and reranking
Retrieval evaluation metrics
Generation evaluation metrics
Production metrics
Architecture selection guidelines
Engineering trade-offs
Common failure modes
Design patterns
Best practices
Evidence-Centric RAG

Audience

This guide is intended for:

AI Engineers
Machine Learning Engineers
Software Architects
RAG Practitioners
Researchers
Students

Resource

📘 RAG Field Guide (PDF)

Research Papers

White Papers

📄 From Uncertainty to Confidence

This site is open source. Improve this page.