Computer Science @ Yonsei Univ.
Undergraduate Intern @ KAIST AI
I am majoring in Computer Science at Yonsei University.
My research goal is ensuring the safety and robustness of embodied systems that interact with the real world.
AI Safety & Security
- Safety of Vision-Language-Action (VLA) systems
- Red-teaming for computer-use and embodied agents
Education
-
Yonsei University. Seoul, South KoreaB.S. Student in Department of Computer ScienceMar. 2021 - Present
Experience
-
KAIST AI, Rise LabAdvised by Prof. Kimin LeeStudent InternJun. 2025 ‑ Present
-
AIM IntelligenceAI Security Measures Research TeamResearch Intern (Research scientist, Mar. - May.)Oct. 2024 ‑ May. 2025 -
Yonsei, MIR LabAdvised by Prof. Youngjae YuStudent InternJan. 2025 ‑ May. 2025
-
Yonsei, YAIAI ClubResearch MemberJul. 2023 ‑ Jun. 2025
-
Yonsei, LikeLionIT Entrepreneurship ClubBackend Member (Organizer, 2023)Mar. 2022 ‑ Dec. 2023
Honors & Awards
-
2nd Prize, Yonsei Prospective Teachers’ Teaching Competition (Presented on ML)2025
-
2nd Prize & People’s Choice Award Yonsei Software Capstone Design Exhibition2024
-
Scholarship Yonsei University Designated Scholarship (TRUTH)2024
-
1st Prize, ABC (AWS Resources Utilization) Hackathon, organized by AWS2024
-
3rd Prize, National Assembly Public Data Utilization Competition2024
-
2nd Prize, AI4Good (GenAI Utilization) Hackathon, organized by the Ministry of Science and ICT & the Innovation Foundation2024
-
3rd Prize, Yonsei Prospective Teachers’ Teaching Competition (Presented on ML)2024
-
2nd Prize, Yonsei Prospective Teachers’ Teaching Competition (Presented on ML)2023
-
3rd Prize, Yonsei Summer Digital Healthcare Bootcamp Advanced Course2023
-
3rd (Silver) Prize, Yonsei GenAI Utilization Competition2023
Selected Publications (view all )
sudo rm ‑rf agentic_security
Sejin Lee*, Jian Kim*, Haon Park, Ashkan Yousefpour, Sangyoon Yu, Min Song (* equal contribution)
Accepted to 63rd Annual Meeting of the Association for Computational Linguistics(ACL) (Volume 6: Industry Track) 2025
SUDO (Screen-based Universal Detox2Tox Offense), a novel attack framework that systematically bypasses refusal-trained safeguards in commercial computer-use agents, such as Claude for Computer Use, is presented.
sudo rm ‑rf agentic_security
Sejin Lee*, Jian Kim*, Haon Park, Ashkan Yousefpour, Sangyoon Yu, Min Song (* equal contribution)
Accepted to 63rd Annual Meeting of the Association for Computational Linguistics(ACL) (Volume 6: Industry Track) 2025
SUDO (Screen-based Universal Detox2Tox Offense), a novel attack framework that systematically bypasses refusal-trained safeguards in commercial computer-use agents, such as Claude for Computer Use, is presented.
All publications
Selected Projects (view all )
EngEagle
Sole Developer
2025
Developed EngEagle, a Chrome extension that streamlines English vocabulary acquisition during web browsing. Implemented instant word lookup and translation with automatic saving to a personal vocabulary list. Built a lightweight review/quiz flow to reinforce saved words with minimal context switching. Led the entire product cycle solo—from problem definition and MVP design to implementation and user-feedback iteration—ranking 1st out of 25 in a peer- and instructor- evaluated course showcase (SW·AI Business Application Design, CAS4151, 2025-2).
EngEagle
Sole Developer
2025
Developed EngEagle, a Chrome extension that streamlines English vocabulary acquisition during web browsing. Implemented instant word lookup and translation with automatic saving to a personal vocabulary list. Built a lightweight review/quiz flow to reinforce saved words with minimal context switching. Led the entire product cycle solo—from problem definition and MVP design to implementation and user-feedback iteration—ranking 1st out of 25 in a peer- and instructor- evaluated course showcase (SW·AI Business Application Design, CAS4151, 2025-2).