HyunJin Kim


Ph.D. Student
HLI Lab, SKKU

Email / Medium / GitHub

About Me

Hello! I am a Ph.D. student at Sungkyunkwan University (SKKU), where I am advised by Prof. JinYeong Bak in the Human Language Intelligence Lab.

I am also in a Ph.D. Collaborator program with Microsoft, where I have worked under supervision of Dr. Young Jin Kim to explore effective offsite-tuning techniques for generative models.

Currently, I am a research intern at Microsoft Research Asia (MSRA) under the guidance of Dr. Xiaoyuan Yi. My work at MSRA focuses on superalignment. I was honored with the Stars of Tomorrow Excellence Award during internship at MSRA.

My research interests lie in analyzing and developing deep learning model architectures (e.g., Transformers, Mixture-of-Experts) and training methods (e.g., Direct Preference Optimization, Instruction Tuning, etc.).

This has led to the development of task-specific architectures such as AsmDepictor and training method like the offsight-tuning, PEMA. Currently, my work focuses on advancing scalable oversight methods to achieve superalignment. For detail, please refer our survey and position.

Education

Preprints and Surveys

  1. Research on Superalignment Should Advance Now with Parallel Optimization of Competence and Conformity
    HyunJin Kim, Xiaoyuan Yi, Jing Yao, Muhua Huang, JinYeong Bak, James Evans, Xing Xie
    [Axriv] 2025.
    PDF

  2. The Road to Artificial SuperIntelligence: A Comprehensive Survey of Superalignment
    HyunJin Kim, Xiaoyuan Yi, Jing Yao, Jianxun Lian, Muhua Huang, Shitong Duan, JinYeong Bak, Xing Xie
    [Axriv] 2024.
    PDF

International Conferences and Journals

  1. PEMA: An Offsite-Tunable Plug-in External Memory Adaptation for Language Models
    HyunJin Kim, Young Jin Kim, JinYeong Bak
    [Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (NAACL)] 2024.
    PDF Code

  2. A Transformer-based Function Symbol Name Inference Model from an Assembly Language for Binary Reversing
    HyunJin Kim, JinYeong Bak, Kyunghyun Cho, and Hyungjoon Koo
    [In the 18th ACM Asia Conference on Computer and Communications Security (ASIACCS)] 2023.
    PDF Code

  3. Associative Knowledge Graph Using Fuzzy Clustering and Min-Max Normalization in Video Contents
    Hyun-Jin Kim, Ji-Won Baek, Kyungyong Chung
    [IEEE Access] 2021.
    Link PDF

Domestic Conferences and Journals

  1. Function Name Prediction using Binary Code with Transformer
    HyunJin Kim and JinYeong Bak
    [The Korean Institute of Information Scientists and Engineers] 2021.

  2. Traffic Knowledge Graph using Associative Document Weight
    Hyun-Jin Kim, Min-Jeong Kim, Ju-Chang Kim, Kyungyong Chung
    [International Conference on Convergence Technology] 2020.

  3. Association Rule based Video Knowledge Extraction using Object Detection Algorithm
    Hyun-Jin Kim, Hye-Jeong Kwon, Ji-Hye Gwon, Kyungyong Chung [Korean Society for Internet Information] 2020.

  4. Data Bias Optimization based Association Reasoning Model for Road Risk Detection.
    Seong-Eun Ryu, Hyun-Jin Kim, Byung-Kook Koo, Hye-Jeong Kwon, Roy C Park, Kyungyong Chung
    [Journal of the Korea Convergence Society] 2020.

Award

Teaching Experience

Talks

Work Experience

Academic Services

Extracurricular Activities

References