| KAIST
avatar

Youngdo Lee

M.S. Candidate

image

lyd0531@kaist.ac.kr


I am an M.S. student in Artificial Intelligence at Korea Advanced Institute of Science and Technology (KAIST), advised by Prof. Jaegul Choo, after receiving my B.S. degrees in Computer Science and Mathematics at the same institution. My broader goal is to develop general-purpose learning agents that acquire skills efficiently from limited experience and act competently in the physical world. My current research focuses on:

  • Sample- & Compute-Efficient Deep Reinforcement Learning: algorithms that let deep RL train stably with larger networks, longer training budgets, and fewer environment interactions.
  • Robot Learning: learning fast, stable, and generalizable controllers for high-dimensional robots, with a focus on whole-body manipulation and humanoid locomotion.



Publications

    • Selected
    • All
  1. RSS
    FlashSAC: Fast and Stable Off-Policy Reinforcement Learning for High-Dimensional Robot Control
    Donghu Kim$^\dagger$, Youngdo Lee$^\dagger$, Minho Park, Kinam Kim, I Made Aswin Nahrendra, Takuma Seno, Sehee Min, Daniel Palenicek, Florian Vogt, Danica Kragic, Jan Peters, Jaegul Choo, Hojoon Lee
    ($\dagger:$ Equal contribution)
    Robotics$:$ Science and Systems (RSS) 2026
    PDF Code Project Page BibTex

  2. ICML
    Hyperspherical Normalization for Scalable Deep Reinforcement Learning
    Hojoon Lee$^\dagger$, Youngdo Lee$^\dagger$, Takuma Seno, Donghu Kim, Peter Stone, Jaegul Choo
    ($\dagger:$ Equal contribution)
    International Conference on Machine Learning (ICML) 2025
    Spotlight Presentation (313/12107 = 2.6%)
    PDF Code Project Page BibTex

  3. RSS
    FlashSAC: Fast and Stable Off-Policy Reinforcement Learning for High-Dimensional Robot Control
    Donghu Kim$^\dagger$, Youngdo Lee$^\dagger$, Minho Park, Kinam Kim, I Made Aswin Nahrendra, Takuma Seno, Sehee Min, Daniel Palenicek, Florian Vogt, Danica Kragic, Jan Peters, Jaegul Choo, Hojoon Lee
    ($\dagger:$ Equal contribution)
    Robotics$:$ Science and Systems (RSS) 2026
    PDF Code Project Page BibTex

  4. RLC
    Unleashing the Architectural Potential of RL in Visual Continuous Control
    Donghu Kim$^\dagger$, Youngdo Lee$^\dagger$, Hojoon Lee, Johan Obando-Ceron, Byungkun Lee, Aaron Courville, Pablo Samuel Castro, Jaegul Choo, Clare Lyle
    ($\dagger:$ Equal contribution)
    Reinforcement Learning Conference (RLC) 2026
    PDF Code Project Page BibTex

  5. CVPR
    Extracting Any Concepts from an Image for Controllable Text-to-Image Generation
    Seunghwan Choi, Jooyeol Yun, Youngdo Lee, Jaegul Choo
    Computer Vision and Pattern Recognition (CVPR) 2026
    PDF

  6. ICML
    Hyperspherical Normalization for Scalable Deep Reinforcement Learning
    Hojoon Lee$^\dagger$, Youngdo Lee$^\dagger$, Takuma Seno, Donghu Kim, Peter Stone, Jaegul Choo
    ($\dagger:$ Equal contribution)
    International Conference on Machine Learning (ICML) 2025
    Spotlight Presentation (313/12107 = 2.6%)
    PDF Code Project Page BibTex

  7. KCC & ICMLW
    Abstraction and Reasoning Challenge with Decision Transformer
    Jaehyun Park, Jaegyun Im, Youngdo Lee, Donghyeon Shin, Sejin Kim, Sundong Kim
    Korea Computer Congress (KCC) 2023
    PDF ICMLW version


Projects

    • Selected
    • All
  1. Real-World Humanoid Locomotion on Uneven Terrain with Sim-to-Real RL
    Developing resilient humanoid locomotion capable of handling irregular and unpredictable terrain using large-scale sim-to-real reinforcement learning.
    KRAFTON AI, Physical AI Team

  2. Real-World Humanoid Locomotion on Uneven Terrain with Sim-to-Real RL
    Developing resilient humanoid locomotion capable of handling irregular and unpredictable terrain using large-scale sim-to-real reinforcement learning.
    KRAFTON AI, Physical AI Team


Work Experience

  1. KRAFTON AI
    [Jun. 2025 - Present]
    Physical AI Team, Research Intern
    Mentor: Kangwook Lee
    TBD

  2. IBS
    [Dec. 2022 - Jun. 2023]
    Data Science Group, Research Intern
    Mentor: Meeyoung Cha and Sundong Kim
    Participated in Abstraction and Reasoning project
    Summary

  3. NCSOFT
    [Jul. 2022 - Aug. 2022]
    Speech AI Lab, Music AI Team, Research Intern
    Mentor: Gyeonghoon Lee
    Developed baseline project for deep learning-based melody generation


Education

  • Korea Advanced Institute of Science and Technology (KAIST)
    [Aug. 2024 - Present]
    M.S. Student in Artificial Intelligence
    • GPA: 4.24 / 4.30

  • Korea Advanced Institute of Science and Technology (KAIST)
    [Mar. 2020 - Aug. 2024]
    B.S. degrees in Computer Science and Mathematics
    • GPA: 4.03 / 4.30 (Summa Cum Laude)



Awards & Honors

  • KAIST Dean's List Award
    [Fall 2022, Spring 2023]
    Awarded to the top 3% of academic performance in KAIST
  • KAIST Presidential Fellowship
    A program that supports a small number of top students to grow as future global leaders in science and technology



Teaching

  • Teaching Assistant
    • CS101: Introduction to Programming, Spring 2024