Lorenzo Cavallaro is a Full Professor of Computer Science at University College London (UCL), specializing in Trustworthy AI for Systems Security. His research focuses on developing learning-based methods that are robust against adversaries by understanding the interplay between program analysis, representations, and machine learning models. His research interests span multiple critical areas in cybersecurity, including adversarial machine learning, malware detection, program analysis, and security evaluation. Cavallaro's work particularly emphasizes the challenges of concept drift in security systems and the development of robust defenses against evolving threats. His research has significant implications for Android security, binary analysis, and memory safety in embedded systems. Analysis of his recent publications (2024-2025) reveals a strong focus on addressing fundamental challenges in ML-based security systems. His work spans malware detection systems that maintain reliability under distribution shifts, adversarial attacks in the problem space, context-driven approaches using LLMs for security applications, and temporal invariance in malware detection. A recurring theme is the critical examination of whether ML-based security systems are truly robust and reliable in real-world scenarios. Cavallaro serves in significant editorial and advisory roles including the NDSS Steering Group (2023-2026), Associate Editor for Computer & Security and ACM TOPS, and Scientific Advisory Board for SERICS. He has been actively involved in program committees for top security conferences including IEEE S&P, USENIX Security, CCS, and NDSS from 2021-2025. He teaches Malware (COMP0060; 2022—ongoing), Research in Information Security (COMP0057; 2021—23), and Computer Security 2 (COMP0055; 2021—ongoing) at UCL, contributing to the next generation of security researchers and practitioners.
