Backend · Infrastructure · Applied AI

Madrid, Spain
Software engineer with 2.5+ years building backend services, real-time distributed systems and CI/CD infrastructure in production, now focused on bringing AI into reliable software. MSc in Computer Science and Technology from UC3M (2026) and a paper published at IberLEF 2026.
November, 2026 — Present
I joined Liferay as a Software Engineer on November 3, 2026, and relocated to Madrid for the role.
March, 2026 — July, 2026
Industry Research Project
As part of my MSc thesis I built a real-time collaborative modelling tool compliant with the SysML v2 standard, in an industry research project at The Reuse Company: an ASP.NET + PostgreSQL backend with HocusPocus (Yjs CRDTs) for conflict-free multi-user editing.
September, 2025 — July, 2026
MSc completed in July 2026 at Universidad Carlos III de Madrid.
September, 2023 — December, 2025
I joined a 3-person team to rebuild from scratch a real-time monitoring platform for nationwide broadcasting infrastructure, a project that had failed over 1.5 years of prior attempts. We delivered it in 6 months and it was adopted by 100+ operations users.
September, 2020 — November, 2025
Studies completed in November 2025. A stage where I developed my passion for software engineering and contributed to projects worth highlighting:
July, 2019
I completed my pre-university studies at Salesianos Stma. Trinidad in the science track.
Software requirements classifier and quality checker (ISO/IEC/IEEE 29148, INCOSE GtWR) that layers deterministic rules with Jev (TypeSafe AI) judgments, rigorously benchmarked against classical ML, a fine-tuned DeBERTa and Claude under a preregistered protocol.
Decentralized end-to-end encrypted messaging platform exploring user-owned identity, distributed data storage, and privacy-first social communication.
AI planning toolkit for temporal multi-agent maze problems in PDDL, featuring OPTIC integration, interactive 3D visualization, graph export, and benchmarking.
Multi-label emotion detection in 16th–17th century Spanish letters. Fine-tuned BETO, mDeBERTa and MarIA with LLM data augmentation and per-label threshold tuning (0.36 Macro F1 on the hidden test). System paper published in the IberLEF 2026 proceedings (CEUR-WS).
Pokémon image classifier built on a ResNet18 convolutional network fine-tuned with PyTorch, packaged as an Electron + React desktop app and shipped alongside a Streamlit demo for quick in-browser inference.
Real-time collaborative diagramming application compliant with the SysML v2 standard, developed as part of my MSc thesis at UC3M. Built with React and Next.js on the frontend, an ASP.NET backend over PostgreSQL, and HocusPocus for conflict-free multi-user editing.