Notes from theDeep Sea
Security
research,
quietly.
Undergraduate at Sangmyung University, studying information security engineering. I spent 2025 in BoB 14, a six-month intensive where I learned to read kernels like systems and attackers like signals.
My work sits at two depths. The shallow: directed fuzzing, deception telemetry, and packet-level attack analysis. The deep: guardrails for small language models deployed in OT and industrial control environments, where a wrong answer can become an operational risk.
In flight,
or already surfaced.
SyzDirect
Directed greybox fuzzing for the Linux kernel. Path-aware seed steering to surface bugs in targeted syscall regions faster than syzkaller's scatter approach.
Honeypot
Deception-oriented telemetry: lure, observe, fingerprint. Attacker-behavior analysis for threat intelligence, built for research deployments.
SLM-based OT
Applied security architecture for OT environments with a small-language-model defensive workflow. Completed during BoB 14 in 2025.