Raúl Fernández del Blanco
Back to the archive

AI

RAG Knowledge Assistant

Generative AI & Semantic Search

A conversational assistant based on Retrieval-Augmented Generation (RAG) for querying documents in natural language, with answers backed by their sources.

Role
Lead frontend developer and sole developer of the RAG backend
Year
2025
Status
Professional project
Category
AI

Context

A generative AI solution that makes it easier to query and retrieve information from documents, combining language models with semantic search techniques.

Approach

Back end, as its sole developer:

  • Ingestion. Automated document preparation and normalization processes.
  • Indexing. Document chunking, embedding generation and incremental indexing.
  • Retrieval. Semantic search and generation of source-backed answers.
  • API. Services in Python and FastAPI, designed around the principle of separation of concerns.

Front end, as lead developer:

  • Interface. A multilingual conversational interface in Astro and React.
  • Architecture. Clean Architecture, Atomic Design and ITCSS to favor reuse and maintainability.

Outcome

I developed the entire RAG backend and took the lead role in the front-end development. The result is a modular solution that combines document retrieval with generative AI.