Kyaw Ye Thu
I'm a final-year undergraduate student at School of Computing of KAIST. My main research interest is about world models, or more broadly generative models, for Physical AI applications such as robotics and autonoumus vehicles. Having background in computer graphics, I'm also intrigued by research in 3D rendering and simulation.
Education
Bachelor of Science Feb. 2027 (Expected)
School of Computing, KAIST
Minor in Business & Technology Management · Semi-minor in Artificial Intelligence
School of Computing, KAIST
Minor in Business & Technology Management · Semi-minor in Artificial Intelligence
Daejeon, KoreaI grew up in a lovely city named Yangon from Myanmar and moved to Korea since 2022 for my tertiary education. Outside my study, I love doing anything related to football, be it playing it physically, playing FC, or watching matches. I also find immense satisfaction in consuming well-crafted stories especially in the form of novels and story-driven video games (Witcher 3, RDR2 for example). I believe stories can tell much about moments in the history and about humanity in ways nonfiction can never capture. Photography is another hobby of mine I recently picked up. Stay tune for my upcoming portfolio! ⚽📖🎮📸
News
August 2026
Joining the GLOW Lab at KAIST AI as an undergraduate research intern.
May 2025
MixCuBe received the Outstanding Paper Award at the C3NLP workshop, NAACL 2025.
Selected Research
Publication · In Progress
Physics-aware Multi-Object 3D Scene Reconstruction (in Progress)
Recently, research in 3D reconstruction shifts from achieving consistency in mere appearance and geometry to attaining physically plausible models of the scene or the object. For this problem, while test-time optimization approaches takes hours to optimize reasonble physical parameters of even a single object, the generalizability of feed-forward approaches is too limited. We are currently striving to overcome the shortcomings of both approaches and provide a simulation pipeline that is easily generalizable.
Publication · C3NLP Workshop @ NAACL 2025 · Outstanding Paper Award
When Tom Eats Kimchi: Evaluating Cultural Bias of Multimodal Large Language Models in Cultural Mixture Contexts
MixCuBe — a cross-cultural VQA benchmark built via a novel image-augmentation pipeline, used to evaluate the cultural bias of SOTA multimodal LLMs in mixed-cultural settings.
Experience
KAISTAug 2026 – Present
Undergrad Research Intern
GLOW Lab (advised by Prof. Seung-Wook Kim)
GLOW Lab (advised by Prof. Seung-Wook Kim)
Jun – Aug 2026C&S
Software Engineering Intern
Drone Software Development
Drone Software Development
KAISTJan – Jun 2026
Student Researcher (URP)
SGVR Lab (advised by Prof. Sung-Eui Yoon)
SGVR Lab (advised by Prof. Sung-Eui Yoon)
Jun – Dec 2025Axinvent
Software Engineering Intern
IoT Development
IoT Development
KAISTAug 2024 – Feb 2025
Undergrad Research Intern
U&I Lab (advised by Prof. Alice Oh)
U&I Lab (advised by Prof. Alice Oh)
Aug 2022 – PresentKAIST
B.Sc Student
School of Computing
School of Computing