Generative Models · Multimodal AI · Robot Learning

Sangmin Lee이상민

Ph.D. Student, School of Computing, KAIST

I am a Ph.D. student in the Scalable Graphics, Vision, and Robotics (SGVR) Lab at KAIST, advised by Prof. Sung-Eui Yoon, with whom I also completed my M.S. I study generative models, particularly diffusion and flow matching, and how to distill them into fast, reliable models. Most recently I have applied this to multi-agent coordination and to consistent text-to-3D generation. I am broadly interested in multimodal models and robot learning.

GitHub Hugging Face

alex6095 [at] kaist.ac.kr l.alex6095 [at] gmail.com Daejeon, Korea

News

Sep 2026
MoSDOT was accepted to NeurIPS 2026.
Sep 2026
Started my Ph.D. at KAIST, continuing in the SGVR Lab with Prof. Sung-Eui Yoon.
Dec 2025
Won 1st place (Grand Prize) at the 2nd Medical AI (MAI) Competition.
Aug 2025
RFDSI was published at CKAIA 2025.

Publications

NeurIPS 2026

MoSDOT: Multi-Agent Coordination via Support-Preserving Distillation

Sangmin Lee, Youngju Na, Chanmi Lee, and Sung-Eui Yoon

Aligns source noise with joint-action modes via semi-discrete optimal transport before teacher training, then distills the teacher into decentralized one-step policies.

Project Paper arXiv BibTeX

Domestic (Korean) Papers

CKAIA 2025

Rectified Flow Distillation via 2nd Order Inversion for Consistent Text-to-3D Generation

Sangmin Lee, Taeyeon Kim, Youngju Na, Jumin Lee, and Sung-Eui Yoon

Shows that stochastic guidance in rectified-flow score distillation gives inconsistent optimization signals, and replaces it with deterministic guidance from second-order ODE inversion for better 3D consistency and finer geometric detail.

Paper

Education

Ph.D. in Computer Science, KAIST

Advisor: Prof. Sung-Eui Yoon

2026 – Present
M.S. in Computer Science, KAIST

Advisor: Prof. Sung-Eui Yoon

2024 – 2026
B.S. in Computer Science, KAIST

Minor in Electrical Engineering · AI concentration

2019 – 2024

Experience

Undergraduate Research Intern, IBS Data Science Group

Protein thermostability prediction with protein language models, with Prof. Meeyoung Cha

Dec. 2022 – Dec. 2023

Teaching

KAIST

Counseling Assistant (CA)

KAIST CA program · academic and career counseling for students

Fall 2026
Operating Systems and Lab (CS30300)

Teaching Assistant

Fall 2025
Artificial Intelligence and Machine Learning (CS50700)

Teaching Assistant

Spring 2025

Awards

1st Place (Grand Prize), 2nd Medical AI (MAI) Competition

2025
1st Place, AI Competition on Biological Research Resources

Minister of Science and ICT Award

2024