Blog
I write about Java, software architecture, design patterns, applied artificial intelligence, RAG, AI agents and software engineering. Through my articles, I share feedback, technical analyzes and concrete approaches to design robust, maintainable and useful applications.
RAG Without a Dedicated Vector Database: Simplifying Infrastructure Without Moving the Problem
An embedded vector store can simplify a RAG system, provided we understand where persistence, availability, and operational responsibilities move.
Observe Before Optimizing
Before optimizing an LLM system, you need to make its decision chain visible so you can understand what it actually does beyond the displayed answer.
Freezing the Behavior of an LLM System Before Evolving It
Changing the model is not enough to guarantee behavioral stability. Characterization tests help protect business invariants and govern the evolution of LLM systems.
RAG in practice: Industrializing unstructured data for AI use cases
Corpus quality, security, governance, indexing, and evaluation: The foundations needed to move from AI POCs to reliable and sustainable use cases.
RAG in Practice : From Retrieval to Business Reasoning
Once the data layer is in place, the real challenge is turning a RAG system that retrieves passages into one that produces structured, useful, and justifiable business answers.
RAG in Practice: Why Everything Starts with Data Engineering
In a real RAG project, answer quality often depends less on the model than on how documents are collected, cleaned, structured, and split. A field perspective based on a project built on a complex business corpus.
RAG with Spring AI: From Naive to Advanced
After building the ingestion pipeline, it's time to query our data. In this article, we compare naive RAG with similarity search to advanced RAG using query rewriting, expansion, and compression with the RetrievalAugmentationAdvisor.
