Kevin Paniagua

Kevin Paniagua Romero is a Full Stack / AI Solutions Engineer at the Computational Intelligence Group (CIG), where he designs and deploys production AI systems on AWS using a fully AI-assisted development approach.

Within CIG, he is the architect behind SoroIA, a Retrieval-Augmented Generation (RAG) conversational assistant for the Museo Sorolla — combining hybrid retrieval, Text-to-SQL and AWS Bedrock LLMs so visitors can explore the museum's collection through natural conversation — and bayes-interpret.com, a cloud platform bringing together algorithms from Master's theses on Bayesian network visualization and interpretation.

Beyond CIG, Kevin has independently built and deployed two further production RAG/agentic systems: edabibliobot, a LangGraph ReAct agent with persistent memory and multi-LLM routing on AWS Bedrock, and bayesium, an academic RAG assistant combining pgvector retrieval, AST-validated Text-to-SQL and formula OCR for scientific documents.

His work spans the full AI product lifecycle — architecture, AWS infrastructure (EC2, RDS, S3, Bedrock), backend/frontend integration and production operations — directing AI-assisted development to deliver and maintain systems used daily by researchers and the public.

Before joining CIG, Kevin worked as a Systems Administrator for UPM's Department of Artificial Intelligence (DIA), and started his career as a Web Developer at Mixin Solutions SL and a Systems/Network Administrator at QUADRAX Consulting. He is currently completing a Degree in Computer Engineering at UPM, building on a Higher Technician in Development of Web Applications (IES Pío Baroja) and a Technician in Microcomputer Systems and Networks (IES Pradolongo).