I am a researcher in AI Security, trustworthy artificial intelligence, and privacy-preserving technologies. I am currently a Research Fellow in the MATS Program, working on inference verification mechanisms to detect model weight exfiltration attacks and advanced security monitoring and control of frontier AI systems.
My research combines modern deep learning with principles from statistical modeling to build systems that are secure, interpretable, and verifiable. I have worked on anomaly and cyberattack detection for industrial and critical infrastructures, behavioral analytics (UEBA), privacy attacks and defenses in federated learning, differential privacy, and explainable deep learning.
Background
I received my PhD in Information and Telecommunications Technologies from the University of Vigo in 2024, awarded Cum Laude with International and Industrial Doctorate distinctions, for the dissertation “Machine Learning Approaches and Explainability for Real-Time Cyberattack Detection”. During my doctoral studies I carried out a research stay at the Department of Computer, Control and Management Engineering (DIAG) of Sapienza University of Rome, working on reinforcement-learning approaches to mitigate denial-of-service attacks in smart grids.
From 2022 to 2026 I led the Data Analytics & AI research group at the Galician Research and Development Center in Advanced Telecommunications (Gradiant) as its Technical Manager, where I set the group’s research agenda in artificial intelligence, cybersecurity, and privacy-enhancing technologies. I acted as principal investigator, consortium coordinator, and work-package leader across 14 competitive European, national, and regional projects, and supervised the researchers and doctoral students carrying out that work. Previously I was a Software Engineer at the Microsoft Canada Development Centre within the Core Data Engineering group, building large-scale telemetry processing for Windows and Azure, and a research assistant at Universidad Carlos III de Madrid. I have also taught as Associate Lecturer in Statistics and Operations Research at the University of Vigo.
Open Science
I develop and maintain neuralGAM, an open-source R and Python implementation of interpretable neural generalized additive models, with more than 250,000 downloads.