Dohyun Lee

Ph.D. Student, KAIST AI

Hi! I am a Ph.D. student at KAIST AI (DAVIAN Lab), advised by Jaegul Choo.

My current research interest lies in how robots can build and use an understanding of the 3D world to manipulate objects and act in real environments. My research centers on two directions: (i) bringing the 3D scene understanding of vision-language models to robot manipulation, and (ii) treating video generation as representation learning for the 3D world and its dynamics, to find what learning objectives yield good representations for world action models.

CV / Email / Google Scholar / LinkedIn / Github

Research Journey

▸2026 –Physical AI
- CoRL 2026first author Procedural Assistance Memory for Proactive Robot
- Feb. 2026 Finished the research internship at KRAFTON AI
▸2024 – 2025Large Language Models
- arXiv 2026 PersonaTrail: Benchmarking Personalized Web Agents through Browsing Trails
- NeurIPS 2025 Delving into Large Language Models for Effective Time-Series Anomaly Detection
- Findings of ACL 2025first author Exploring In-context Example Generation for Machine Translation
- ACL 2025 Revisiting LLMs as Zero-Shot Time Series Forecasters: Small Noise Can Break Large Models
- ACL 2025 Opt-Out: Investigating Entity-Level Unlearning for Large Language Models via Optimal Transport
- Jan. 2025 PUBG Ally shown at CES 2025 in collaboration with NVIDIA
- Nov. 2024 Joined KRAFTON AI as a research intern, AI Companion Team
- arXiv 2024 Breaking Chains: Unraveling the Links in Multi-Hop Knowledge Unlearning
- COLM 2024 PairEval: Open-domain Dialogue Evaluation with Pairwise Comparison
- Findings of ACL 2024first author Protecting Privacy Through Approximating Optimal Parameters for Sequence Unlearning in Language Models
- Aug. 2023 Started the integrated M.S./Ph.D. at KAIST AI, DAVIAN Lab

Publications

Procedural Assistance Memory for Proactive Robot
D. Lee*, M. Park*, J. Park, D. Hwang, J. Hyung, S. Hwang, B. Lee, and J. Choo
CoRL 2026.
paper / project page

PersonaTrail: Benchmarking Personalized Web Agents through Browsing Trails
S. Yang, C. Ki, D. Lee, J. Choo, and C. Park
arXiv 2026.
paper

Delving into Large Language Models for Effective Time-Series Anomaly Detection
J. Park, K. Jung, D. Lee, H. Lee, D. Gwak, C. Park, J. Choo, and J. Cho
NeurIPS 2025.
paper

Opt-Out: Investigating Entity-Level Unlearning for Large Language Models via Optimal Transport
M. Choi, D. Rim, D. Lee, and J. Choo
ACL 2025.
paper

Revisiting LLMs as Zero-Shot Time Series Forecasters: Small Noise Can Break Large Models
J. Park, H. Lee, D. Lee, D. Gwak, and J. Choo
ACL 2025.
paper

Exploring In-context Example Generation for Machine Translation
D. Lee, S. Lee, C. Yang, Y. Baek, and J. Choo
Findings of ACL 2025.
paper

Breaking Chains: Unraveling the Links in Multi-Hop Knowledge Unlearning
M. Choi, C. Park, D. Lee, and J. Choo
arXiv 2024.
paper

PairEval: Open-domain Dialogue Evaluation with Pairwise Comparison
C. Park, M. Choi, D. Lee, and J. Choo
COLM 2024.
paper

Protecting Privacy Through Approximating Optimal Parameters for Sequence Unlearning in Language Models
D. Lee*, D. Rim*, M. Choi, and J. Choo
Findings of ACL 2024.
paper

Work Experience

KRAFTON AI Nov. 2024 – Feb. 2026

Research Intern, AI Companion Team Seoul, Korea

Built an AI co-playable character (CPC) for PUBG Ally on top of LLM/VLM-based agent systems. demo
PUBG Ally was showcased at CES 2025 in collaboration with NVIDIA, and beta-launched in PUBG: BATTLEGROUNDS.

Hyundai Motor Company Jul. 2022 – Aug. 2022

Undergraduate NLP Researcher Seoul, Korea

Education

KAIST Sep. 2023 – Present

Integrated M.S./Ph.D. in Artificial Intelligence (Advisor: Prof. Jaegul Choo) Seongnam, Korea

Kyung Hee University Mar. 2017 – Aug. 2023

B.S. in Mechanical Engineering and Software Engineering Yongin, Korea

4 Academic Excellence Scholarship, 2 Academic Excellence Award