معرفی
Stjepan Picek serves as an Associate Professor at Delft University of Technology in the Intelligent Systems school and maintains a secondary affiliation with Radboud University's Digital Security Group in Nijmegen, The Netherlands. His research bridges cryptographic theory with practical machine learning applications, particularly focusing on security vulnerabilities in modern computing systems.
Dr. Picek's research spans multiple critical areas in cybersecurity, with particular emphasis on deep learning-based side-channel analysis where he has published extensively on improving attack methodologies and countermeasures. His work on cryptographic Boolean functions explores evolutionary computation approaches to design functions with optimal cryptographic properties. More recently, his research has expanded to AI security, investigating backdoor attacks, membership inference vulnerabilities, and covert channels in large language models.
Analysis of his recent publications reveals a clear trajectory toward integrating traditional cryptographic techniques with modern machine learning paradigms. His work demonstrates consistent innovation across three main axes: (1) advancing side-channel analysis through deep learning innovations, (2) developing evolutionary methods for cryptographic primitive design, and (3) identifying and mitigating security vulnerabilities in emerging AI systems. This multidisciplinary approach positions his research at the critical intersection of theoretical cryptography and practical security implementation.
Dr. Picek maintains extensive collaborations with researchers across Europe, particularly with Domagoj Jakobovic (104 co-publications), Luca Mariot (56), Lejla Batina (56), and Guilherme Perin (56), indicating a well-established research network and leadership in his field. His publication record shows remarkable productivity with over 380 indexed publications, including 63 in 2023, 53 in 2024, and 24 in 2025, demonstrating sustained research momentum.
