E-mail : lyh4186@naver.com
๐ย Research interest:
System Security, Security for AI
I am a Cybersecurity Engineer at AutoCrypt, focusing on vehicle and system security. My work bridges practical automotive vulnerability research and academic research on microarchitectural side-channel attacks.
I hold an M.S. from the Graduate School of Information Security at Korea University, where I was advised by Professor Junbeom Hur. My research focused on system security, particularly side-channel attacks and their mitigation across processor architectures and AI systems.
During my graduate studies, I demonstrated the first software-based power side-channel attack against Multi-Exit Neural Networks in GPU server environments. Using unprivileged interfaces such as nvidia-smi, I showed that architectural information can be inferred even under delay-based defenses like TimeGuard. This work introduced the novel concept of artificially added delays as an attack vector, expanding power side-channel analysis beyond traditional computation-induced variations.
In parallel, I developed a machine learningโbased detection model for Prefetch Side-Channel Attacks using Hardware Performance Counters and analyzed Meltdown-based data leakage paths through CPU cache and memory hierarchies.
At AutoCrypt, I extend this expertise to the automotive domain, analyzing real-vehicle vulnerabilities and communication protocols such as CAN, UDS, and DoIP, as well as BLE and USB interfaces. My focus is on identifying and mitigating security weaknesses in connected and autonomous vehicle systems by integrating low-level system security insights into real-world vehicular environments.
2023.09 ~ 2025.08
2017.03 ~ 2023.08
AutoCrypt (2025.08 ~ Present / ์ ๊ท์ง)
**TรV SรD ์ฝ๋ฆฌ์ (**2024.01 ~ 2024.02 / Internship)
**์ธ๋ํฌ๋ ์คํธ (**2021.07 ~ 2021.08 / Internship)