📚 Featured Research
Cross-architecture Linux malware analysis framework via lightweight QEMU emulation and eBPF monitoring. Achieves 94.7% detection accuracy across 11 CPU architectures.
Security kernel approach for AI agents with evidence-carrying actions and capability-ledger access control. Addressing prompt injection and privilege escalation risks.
💻 Featured Projects
Malware analysis framework supporting multi-architecture Linux binaries via QEMU user-mode emulation and eBPF monitoring. Part of MirrorShield research.
Rewriting OpenCV's resize function using SDIM and OpenMP for faster performance. Capable of handling large images efficiently with high performance.
Scientific research literature database built with modern web technologies. Supports paper management, search, and academic collaboration features.