Dongjun Kim


I am a postdoctoral scholar in the Prof. Stefano Ermon's group at Stanford University. I am interested in generative AI for computer vision. My goal is to develop a simple yet effective and trustworthy new framework for the generation task that can broadly impact our real life.


Previously, I worked with Sony AI and got my Ph.D at KAIST in 2023, advised by Prof. Il-Chul Moon. 

Selected Publications (Full List of Publications)

PaGoDA: Progressive Growing of a One-Step Generator from a Low-Resolution Diffusion Teacher [pdf]

Consistency Trajectory Models: Learning Probability Flow ODE Trajectory of Diffusion [pdf]

Ph.D. Thesis

A Study on the Score-based Diffusion Model for Improved Training, Flexible Inference, and Efficient Sampling [pdf]