About
Dongho Seo (서동호) · Undergraduate Researcher, SYMPLE Lab, GIST
I am a senior undergraduate in Electrical Engineering and Computer Science at GIST and a research intern at SYMPLE Lab, advised by Prof. Sangjin Kim.
I work on KV cache optimization for long-context LLM inference, and I am expanding toward on-device inference and NPU design.
The latest version of this profile lives at dongho-seo.me.
Research Interests
- KV cache optimization for long-context LLM inference
- On-device LLM inference
- NPU design for generative AI workloads
Publications
Dual-QK: Sharp Queries and Flat Keys for Prunable 2-bit KV Caches
Sunjoo Whang, Jungjun Oh, Minsung Kim, Dongho Seo, Jisu Shin, Gregory Kielian, Hoi-Jun Yoo, Sangjin Kim
arXiv preprint, 2026 · arXiv · PDF
Research Experience
SYMPLE Lab, GIST · Jul 2026 – Present
Undergraduate Research Intern · Advisor: Prof. Sangjin Kim
Efficient LLM inference: KV cache quantization and pruning.
Education
Gwangju Institute of Science and Technology (GIST) · Mar 2020 – Aug 2027 (exp.)
B.S. in Electrical Engineering and Computer Science
University of California, Berkeley · Summer 2023
Berkeley Summer Sessions
Projects
Gijol: Graduation Requirement Checker · Apr 2022 – Present
Full-stack developer. Checks graduation requirements from uploaded transcripts. Used by about 51% of GIST undergraduates.
Website · Retrospective
Mefeto: AI-assisted Legislative Information Platform · Apr – Aug 2023
Frontend developer. Summarizes National Assembly bills with GPT and visualizes related issues.
Honors & Awards
- Excellence Award, 9th Gwangju Public Data Startup Competition (2023)
- Special Prize, AIFactory ChatGPT Microsoft Azurethon (2023)
Service
- GIST Student Association, InfoTeam · Full-stack developer · Mar 2026 – Present
- Republic of Korea Air Force · Information Technology Specialist (mandatory military service) · Mar 2024 – Dec 2025