About me

I’m a second-year Ph.D student of Efficient Computing Lab in the Department of Computer Science & Engineering at POSTECH, advised by Prof. Eunhyeok Park. Before joining POSTECH, I completed my B.S. in Department of Computer Science & Engineering in Kyung Hee University.

I’m currently focusing on Efficient AI, particularly in enhancing Memory Efficiency and Computation Acceleration during model training and inference of various models (Vision, LLM, Video Generation, etc.) via Quantization and Low-rank Approximation.

Research Keywords

• Fast and Memory Efficient Training (A01, C03)
• Fast and Memory Efficient Inference (C01, C02)
• Parameter Efficient Fine-tuning of LLMs (U01, P03)
• CUDA Kernel optimization (P01, P02, P03, C01, C03)
• Fast Sampling of Video Generation Diffusion Models

News

• [Nov. 03, 2025] I’ve joined AMD as a Research Associate.
• [Oct. 21, 2025] I’ve been selected as a Winner of Qualcomm Innovation Fellowship Korea.
• [Oct. 01, 2025] I’ve been selected for the BK21 Outstanding Graduate Student International Training Scholarship by Ministry of Science and ICT, South Korea.
• [Mar. 02, 2025] 1 paper has been accepted to ICML 2025.
• [Feb. 25, 2025] 1 paper has been accepted to CVPR 2025.
• [Oct. 28, 2024] 1 paper has been accepted to WACV 2025.

Publications

• [C03] HOT: Hadamard-based Optimized Training
  Seonggon Kim, Juncheol Shin, Seung-taek Woo, Eunhyeok Park
  Computer Vision and Pattern Recognition (CVPR 2025), Nashville.

• [C02] Merge-Friendly Post-Training Quantization for Multi-Target Domain Adaptation
  Juncheol Shin, Minsang Seok, Seonggon Kim, Eunhyeok Park
  International Conference on Machine Learning (ICML 2025), Vancouver.

• [C01] PTQ4VM: Post-training Quantization for Visual Mamba
  Younghyun Cho*, Changhun Lee*, Seonggon Kim, Eunhyeok Park
  Winter Conference on Applications of Computer Vision (WACV 2025 Oral), Tucson.

• [U01] HoLA: Overcoming the full-finetuning with Hadamard-oriented LoRA
  Seonggon Kim, Taehyeon Kim, Byeori Kim, Eunhyeok Park
  Neural Information Processing Systems (NeurIPS 2025, Under review), San Diego.

• [A01] HLQ: Fast and Efficient Backpropagation via Hadamard Low-rank Quantization
  Seonggon Kim, Eunhyeok Park
  arXiv 2406.

Project

• [P03] Fast and Memory-efficient training on Extreme environment, Jul. 2024 – Current
  National AI Research Lab of Korea
  • Conducted research on memory-efficient training for vision models.
  • Prototype development of an optimized CUDA kernel for memory-efficient training.

• [P02] GEMV Accelerator for LLM inference on Intel Gaudi-2, Jun. 2024 - Jun. 2025
  Naver & Intel Joint Research Center
  • Conducted research on LLM’s fast inference on Intel Gaudi-2 architecture.
  • Implemented custom GEMV kernel for Gaudi with TPC-C language.
  • Transplanted LUT Quantization from CUDA to Gaudi TPC.

• [P01] Solutions for Self-supervised training on Edge Device, Jun. 2023 – Current
  Ministry of Science and ICT of Korea
  • Conducted research on fast fine-tuning on Edge device.
  • Designed an efficient fine-tuning algorithm with stochastic quantization.
  • Implemented custom CUDA kernel for fast fine-tuning.

Experience

• Research Associate, Nov. 2025 - Current
  AMD, Longmont, CO, USA

• Software Engineer Intern, Jul. 2022 - Feb. 2023
  Spirent Communications, San Jose, CA, USA

• Software Engineer Intern, Feb. 2022 - Jun. 2022
  Common Computer, Seoul, Korea

• Research Intern, Mar. 2021 - Dec. 2021
  SI Analytics, Daejeon, Korea

Awards & Honors

• Qualcomm Innovation Fellowship Korea, Winner, Oct. 2025

• BK21 Outstanding Graduate Student International Training Scholarship, Selected Graduate Student, Oct. 2025

• ETHDenver 2022 Blockchain Hackathon NFT project, 3rd Prize
  SporkDAO, Feb. 2022

• CVPR 2021 Earthvision workshop, Land Cover Classification Challenge, 5th Prize
  CVPR, Jun. 2021

Education

• M.S/Ph.D. in Computer Science and Engineering, POSTECH
  Sep. 2023 - Present

• B.S. in Computer Science and Engineering, Kyung Hee University
  Feb. 2017 - Aug. 2023

Teaching Experience

• Teaching Assistant, Mar. 2025 - June. 2025
  CSED311: Computer Architecture, POSTECH