Seohui Bae 배서희

I'm a research scientist at LG AI Research in South Korea.

At LG AI Research, I'm working on AI systems that model the world, make decisions, and adapt under uncertainty. I completed my bachelor's and master's studies at KAIST, where I was fortunate to be advised by Prof. Eunho Yang.

Email  /  Google Scholar  /  LinkedIn

News

• (26.05) I will attend ICML 2026 in Seoul 🇰🇷
• (26.02) 1 paper accepted to CVPR 2026. See you all in Denver 🇺🇸
• (26.01) 1 paper accepted to ICASSP 2026.
• (25.07) I will be attending ICML 2025 this July. See you in Vancouver 🇨🇦

Research

My research focuses on building AI agents that can model, decide, and adapt in complex environments. I study how agents can use executable dynamics, grounded environments, and model-based exploration to make robust decisions under uncertainty. My work spans three complementary directions:

• Agent Decision-Making: RL and planning methods for intelligent agents in dynamic environments, with a focus on exploration and information-seeking behavior. [P2, P3, C1]
• World Models & Agent Environments: Building grounded, scalable environments and learning predictive models of dynamics for simulation and model-based reasoning. [P3, P4]
• Robust Learning under Distribution Shift: Learning algorithms for reliable generalization under changing distribution. [P1]

I regularly contribute to academic publications and collaborative research projects. I’m especially interested in bridging industrial challenges with generalizable solutions in: RL post-training, world models, and environment scaling.

Selected Publications

(* equal contribution; † corresponding author). For the full list, see Google Scholar.

[C#] conferences; [P#] preprints/under review

• [P4] PCBWorld: A Benchmark Environment for Engine-Grounded PCB Design Automation [project]
  Hyungseok Song*, Junseok Park*, Won-Seok Choi*, Seohui Bae, Han-Seul Jeong, Youngjoon Park†, Soonyoung Lee†
  under review
  KDD Workshop on Evaluation and Trustworthiness of Agentic AI, 2026
• [P3] Code-to-Explore: Verifiable Programs for OOD Exploration in LLM Agents [pdf]
  Seohui Bae, Whiyoung Jung, Youngjoon Park, Soonyoung Lee
  under review
  ICML Workshop on Reinforcement Learning on World Feedback, 2026
• [P2] Align as Act: Innovations-Based Reward Decomposition for LLM Agents [pdf]
  Sojeong Rhee*, Seohui Bae*, Whiyoung Jung, Woohyung Lim, Soonyoung Lee, Youngchul Sung
  preprint, 2026
• [C1] Align While Search: Belief-Guided Exploratory Inference for World-Grounded Embodied Agents [pdf]
  Seohui Bae, Jeonghye Kim, Youngchul Sung, Woohyung Lim
  Conference on Computer Vision and Pattern Recognition (CVPR), 2026
  ICML Workshop on Exploration in AI Today, 2025
• [P1] Geometry-Aware Normalization for Imbalanced Time-series Forecasting [pdf]
  Seohui Bae, Junhyeok Kang, Jun Seo, Soyeon Park, Wonbin Ahn, Soonyoung Lee
  under review (journal), 2025

Projects

Current @ LG AI Research

• RL for Foundation Model Agents
  • Post-training methods for LLM-based agents in industrial decision settings.
• Industrial Optimization
  • RL for sequential decision-making in manufacturing.
  • Engine-grounded benchmark for electronic design automation.

Past @ LG AI Research

• EXAONE-Futurecast
• Demand Forecasting

Education

M.S. in Graduate School of Artificial Intelligence, Mar 2020–Feb 2022

• Korea Advanced Institute of Science and Technology (KAIST)
• Machine Learning and Intelligence Lab
• Advisor: Eunho Yang

B.S. in Biological Science, Computer Science (minor), Mar 2015–Feb 2020

• Korea Advanced Institute of Science and Technology (KAIST)

Korea Science Academy of KAIST, Mar 2012–Feb 2015

Academic Service

Conference Reviewer

• Main: ICML (Top Reviewer, 2026), NeurIPS, ICLR, AAAI, AISTATS
• Workshops: AAAI, ICLR, ICML

Journal Reviewer

• ACM Computing Surveys

Last date of update: 2026-05-16 / template