Profile

Intro

I am a PhD student in the Computer Vision Lab at Yonsei University, advised by Prof. Bumsub Ham. My research broadly explores how to make deep learning models more efficient and effective. My earlier work focused on computational efficiency, particularly model quantization and neural architecture search (NAS). More recently, I have extended this direction to large-scale generative models, from accelerating diffusion transformers to improving the generation of diffusion-based large language models. I am also interested in agentic AI, with the goal of building intelligent and autonomous systems that are efficient, scalable, and practical for real-world use.


Education


Publications

(*: equal contribution)


Awards

Silver Prize, 32nd Samsung Humantech Paper Award, 2026

Experience

Researcher at Articron (Aug. 2023 – Present)
  • Developed quantization and pruning algorithms for CIM-based NPUs across diverse model families, including CNNs, LLMs, and audio models, in close collaboration with hardware teams.
  • Built deep learning methods for circuit performance prediction and design optimization.

Patents


Services