Taeyoon Kim

Hi! I am a undergraduate student at Seoul National University, majoring in Computer Science and Engineering and Mathematical Sciences. Also I am currently an intern in Cognitive Learning for Vision and Robotics (CLVR) Lab advised by Prof. Joseph J. Lim.

Previously, I worked as a research intern at SNU CVLab under professor Bohyung Han, where I worked on diffusion models, especially on inference steering techniques on video diffusion models.

For more information, please check my CV!

Email  /  CV  /  LinkedIn  /  Github

profile photo

Research Goal

My long-term goal is to build truly intelligent systems.
I am currently focused on robotics and reinforcement learning, as they enable the use of continuous rich real-world data
to ground intelligence in common sense and physical understanding. My research interests include:

  • Continual lifetime learning with feedback: Reinforcement Learning, Meta Learning
  • Utilizing large-scale, multi-modal data: SSL (LLMs, diffusion models), Multi-modal learning
  • Data-driven learning: Algorithms that can scale
  • Efficient, common-sense learning: Neuroscience-inspired deep learning
I'm currently working on Self-Improving Physical AI.

Research Projects
Motion LoRA animation Motion LoRA: Learn motion using Low-Rank Adaptation
fal.ai Research Grant, 2024
code

Implemented a Low-Rank Adaptation (LoRA) training algorithm to enhance the Stable Video Diffusion model, enabling video generation aligned with motion patterns from limited video datasets.

SV3D Fine tuning animation SV3D Fine tuning
fal.ai Research Grant, 2024
code

Open-sourced SV3D training code, the first release of its kind worldwide


Template from jonbarron.