yiyue@skku
0x2000A

.rodata · NUL-terminated strings · hover a byte to light its character

Yiyue Zhang

I take software apart to find out what it actually does — not what its source claims.

Integrated M.S.–Ph.D. student at SecAI Lab, Sungkyunkwan University. Binary analysis and reverse engineering.

shdr

.shstrtab

$ readelf -SW yiyue

Section headers — addresses ascend downward
[Nr]NameTypeAddressOffSizeFlg
[ 6].dynsymDYNSYM00000000000003c00003c00000f0A
[11].pltPROGBITS0000000000001020001020000030AX
[13].textPROGBITS00000000000010600010600003a2AX
[15].rodataPROGBITS00000000000020000020000000b8A
[16].noteNOTE00000000000020b80020b8000048A
[19].init_arrayINIT_ARRAY0000000000003d90002d90000018WA
[25].commentPROGBITS000000000000000000303200002bMS
0x0000MS

.comment

who produced this

About

I'm an integrated M.S.–Ph.D. student in the Department of Software at Sungkyunkwan University, working in SecAI Lab under Prof. Hyungjoon (Kevin) Koo. I joined the lab in Fall 2024 and begin the Ph.D. stage in September 2026.

Most of what I do involves reading code that was never meant to be read: stripped binaries, aggressively optimized builds, deliberately obfuscated targets. I'm interested in the seam where automated analysis gives up and someone still has to sit down with a disassembler — and in how much of that gap we can actually close.

0x1060AX

.text

what actually executes

Research

SecAI Lab studies practical security across both offensive and defensive dimensions — understanding attacks such as code injection and code reuse, designing defenses such as attack surface reduction and moving target defense, and examining security-relevant artifacts. To reason about code semantics, the lab works through the stages an executable binary passes through:

  1. code generation

    transforming high-level source into machine code

  2. obfuscation

    techniques that deliberately hinder analysis

  3. static & dynamic analysis

    reasoning about structural and behavioral properties

  4. bug discovery

    inspecting program behavior to uncover vulnerabilities

  5. patching

    fixing security flaws

  6. decompilation

    reconstructing higher-level representations to support reversing

The direction I find most compelling is the one the lab frames as an AI-driven pipeline: unifying stages that have traditionally been treated as separate problems, so that analysis of benign and malicious software becomes both scalable and resilient. The lab's second track — securing AI models themselves across the lifecycle, against poisoning, backdoors, evasion, extraction, inversion, and membership inference — sits on the other side of the same question.

0x1020AX

.plt

calls out to the world

Contact

Email
yiyue AT g.skku.edu
GitHub
github.com/Yii-99
Lab
SecAI Lab, SKKU