I'm completing an Applied Artificial Intelligence degree while building computer vision, explainable AI, and full-stack systems through research and industry projects.
At the University of Jaén, I developed an end-to-end fake news detection platform on the LIAR-2 benchmark, using BERT, RoBERTa, DistilBERT, and a self-crafted multi-task learning architecture that achieved over 90% accuracy. I also benchmarked LIME, SHAP, DeepSHAP, and DeepLIFT for real-time explainability and integrated Alastria Blockchain smart contracts for EU AI Act auditability.
At Ubotica and Vision AgeVFX, I optimized deep learning models for instrument OCR and automatic meter reading, then built a YOLOv11 screen-detection workflow that achieved 78% mAP50. I also created a real-time video pipeline for screen tracking and dynamic VFX replacement.
I've built practical AI applications including gesture volume control, a self-driving car simulation, multiple sclerosis prediction, and autism screening analysis. My work combines Python, PyTorch, OpenCV, transformers, MLOps, and data analysis to turn AI models into usable systems.
