Yeongtak Oh, a Ph.D. candidate in the Data Science & AI Laboratory (advised by Professor Sungroh Yoon) of Seoul National University’s Department of Electrical and Computer Engineering, has been selected as a Winner in the Qualcomm Innovation Fellowship Korea (QIFK) 2025, announced at the final event held on November 27.

QIFK is a scholarship and research support program organized by Qualcomm for graduate students in Korea, designed to identify and support outstanding research in AI through dedicated funding and scholarships. More than 120 submissions were received in this year’s edition, from which 15 were ultimately selected as Winners.

Oh’s research proposes the world’s first post-training methodology for personalizing multimodal large language models (LLMs) based on reinforcement learning. The approach demonstrates significantly superior generalization performance compared to existing post-training techniques, particularly in image captioning tasks involving complex scenes where multiple concepts coexist within a single image.

Image: Overview of the proposed methodology

Image: Ph.D. candidate Yeongtak Oh

Image: Group photo of the Winners

Image: Ph.D. candidate Yeongtak Oh delivering his presentation

Source: https://ece.snu.ac.kr/ece/news?md=v&bbsidx=57099

Translated by: Changhoon Kang, English Editor of the Department of Electrical and Computer Engineering, changhoon27@snu.ac.kr