Geewook Kim is a team manager (leader) and applied research scientist at NAVER Cloud, where he leads applied research and engineering for HyperCLOVA X's multimodal models (Vision, Video, and Omni) and drives omni-modal LLM post-training. He received his Ph.D. in Artificial Intelligence from KAIST in 2026, advised by Prof. Minjoon Seo, and his B.Eng. and M.Informatics degrees from Kyoto University in 2018 and 2020, advised by Prof. Hidetoshi Shimodaira. His research builds resource-efficient multimodal AI: models that perceive across text, image, audio, and video under real-world compute and data limits, together with benchmarks and evaluation methods that probe what such models actually understand. He initiated and led Donut, a widely adopted open-source framework for OCR-free document understanding, and has published 25 peer-reviewed international conference papers, 14 of them as first or co-first author, at venues including ICML, ICLR, AAAI, ECCV, ICCV, ACL, EMNLP, and NAACL.