01About
I build intelligent systems at the intersection of machine learning research and production-grade engineering. My recent work spans diffusion models, computer vision, and full-stack web applications. I care about clarity, rigor, and craftsmanship.
- Role
- Student Researcher
- Focus
- ML / CV / Systems
- Languages
- Korean, English
- Status
- Open to opportunities
02Education
-
2021 — Present
KAIST — Korea Advanced Institute of Science & Technology
B.S. in Computer Science. Coursework in Algorithms, Machine Learning, Computer Vision, Diffusion Models.
-
2018 — 2021
High School
Graduated from Daewon High School.
03Selected Projects
A curated set of work across research, applied ML, and engineering. Click any banner for a detailed write-up.
-
Read →Variable-sized K-ary Tree Modeling via Discrete Flow Matching
A discrete flow-matching framework that generates rooted K-ary trees of arbitrary size while preserving structural rules — parent–child consistency and acyclicity — through edit-based (insert / delete / substitute) flow on tree slots.
-
Read →Simpsonization — Identity-Preserving Face Stylization
A diffusion-based pipeline that turns real faces into Simpson characters while preserving identity and scene context — combining a Simpson-style LoRA, a 9-channel inpainting UNet, and Canny ControlNet for facial structure.
04Skills
Languages
Python, C/C++, JavaScript, TypeScript, Java, SQL
Machine Learning
PyTorch, NumPy, Diffusion Models, Computer Vision, Deep Learning
Web & Systems
Flask, React, Node.js, Linux, Git
Tools
VS Code, Jupyter, Anaconda, Android Studio
05Contact
For collaboration, research opportunities, or general inquiries.
- Emaildmsdmswns@kaist.ac.kr
- GitHubgithub.com/k1shooter
- LinkedInlinkedin.com/in/seungjun-kim-816a59375