Jaehyun Kang

Research Engineer @ RLWRLD

profile.jpg

I am a research engineer at RLWRLD. I received my M.S. and B.S. in Computer Science from Yonsei University, where I was advised by Prof. Seon Joo Kim in CIPLAB. I also had the fortunate opportunity to collaborate with Prof. Youngwoon Lee, which led to our work on learning cross-embodiment skills from human videos.

My research centers on dexterous manipulation for long-horizon tasks. Skillful control alone is not enough to solve them: a robot must also reason about its goal and act in a context-aware way, adapting to the scene and the objects involved. This leads me to focus on designing robot intelligence across levels of abstraction, where high-level reasoning guides low-level dexterous control to solve problems that neither could handle alone.

News

May 07, 2026 RLDX-1 Technical Report is out! Check out the report, code, and project page.
Aug 01, 2025 UniSkill is accepted to CoRL 2025.
Jun 16, 2025 Joined RLWRLD as a Research Engineer.
May 13, 2025 UniSkill is out. Check out our paper and project website.

Publications

(* indicates equal contributions)

  1. RLDX-1 Technical Report thumbnail
    RLDX-1 Technical Report
    Dongyoung Kim et al.
    arXiv 2026
  2. UniSkill: Imitating Human Videos via Cross-Embodiment Skill Representations thumbnail
    UniSkill: Imitating Human Videos via Cross-Embodiment Skill Representations
    CoRL 2025
  3. VISAGE: Video Instance Segmentation with Appearance-Guided Enhancement thumbnail
    VISAGE: Video Instance Segmentation with Appearance-Guided Enhancement
    ECCV 2024
  4. Leveraging Image Augmentation for Object Manipulation: Towards Interpretable Controllability in Object-Centric Learning thumbnail
    Leveraging Image Augmentation for Object Manipulation: Towards Interpretable Controllability in Object-Centric Learning
    Jinwoo Kim* , Janghyuk Choi* , Jaehyun Kang , Changyeon Lee , Ho-Jin Choi , Seon Joo Kim
    CVPR Workshop 2024