Welcome to OSI LAB!

We are the Optimization and Statistical Inference Lab of KAIST AI, located in the Seoul Hongneung Campus. We tackle a wide range of problems from RL theory to generative AI. If you are interested in joining us, please refer to our application page.

Group Photo
Latest News
10 papers accepted to NeurIPS 2026!
  • DualDrift: Combining Forward and Reverse Drifts for One-Step Generative Modeling
  • MolHIT: Advancing Molecular-Graph Generation with Hierarchical Discrete Diffusion Models
  • Stage-wise Attention-Guided Region Sequencing for Adversarial Attacks on Large Vision-Language Models
  • Instance-Optimal Estimation with Multiple LLM Judges on a Budget
  • Provably Efficient Regularized Online RLHF with Generalized Bilinear Preferences
  • Looking Through the Mirror: Minimax-Optimal Regularized Regrets in Online Learning and Bandits
  • Variational Monte Carlo for Quantum Excited States via Nested Low-Rank Approximation
  • BASTION: Budget-Aware Speculative Decoding with Tree-structured Block Diffusion Drafting
  • Temporal Alignment Guidance: On-manifold Sampling in Diffusion Models
  • LLM-as-a-Tutor: Policy-Aware Prompt Adaptation for Non-Verifiable RL
2 papers at CKAIA 2026 (한국인공지능학회) awarded Distinguished Paper Award (우수논문상)!
1 paper at ICML 2026 CTB Workshop awarded Best Paper (Long Paper Track)!