Professor
Seungjun Shin (신승준)
Assistant Professor | AI-Native Systems Lab (ANSL)
Division of Artificial Intelligence Engineering, Sookmyung Women's University
Email: seungjun@sookmyung.ac.kr
Office: 수련교수회관 314호, Sookmyung Women's University
Education
| Organization | Degree |
|---|---|
| POSTECH | Ph.D. in Electrical Engineering (Integrated M.S./Ph.D.) Advisor: Woo-Jin Song |
| POSTECH | B.S. in Electrical Engineering |
Experience
| Organization | Role |
|---|---|
| Sookmyung Women’s University | Assistant Professor |
| TwelveLabs | Research Scientist (Video Language model) |
| Samsung Electronics (SAIT / S.LSI) | AI Researcher (Autonomous Driving / Neural Codec / AI System) |
| LG Display | AI Researcher (AI Algorithm Task) |
Selected Publications
- S. Shin, J. Oh, and D. Oh, “OrthoRank: Token Selection via Sink Token Orthogonality for Efficient LLM Inference,” ICML, 2025.
- S. Shin, S. Kim, and D. Oh, “Efficient Neural Video Representation with Temporally Coherent Modulation,” ECCV, 2024. Oral
Honors & Awards
- Samsung Best Paper Award: Bronze Prize (2025), Silver Prize (2020)
- Continual Learning Challenge for CVPR Workshop: 1st Place (Supervised Learning Track, 2021)
- Dongbu Cultural Foundation Scholarship (2010, 2011)
Patents
[Granted]
- US 12,536,765: Method and apparatus for extracting the region of interest (2026)
- US 12,229,965: Image segmentation method and device (2025)
- KR 102072757B1: Heterogeneous sensor information fusion using deep learning (2020)
[Selected Applications]
- LLM Optimization: Memory reduction, Model compression, and Cache optimization (US 19/222,336, KR 10-2025-0051613, KR 10-2025-0047503)
- Position Encoding: Electronic device augmenting position information of tokens (US 19/262,902)
- Neural Representation: Implicit neural video representation apparatus (US 18/679,066)