Two UNIST graduate students have been selected for the  2026 Presidential Science Scholarship for research in green hydrogen production and artificial intelligence (AI).

EunSong Jeon, a Ph.D. student in the Graduate School of Carbon Neutrality, is developing a photoelectrochemical system that produces green hydrogen using sunlight. Her work combines an organic semiconductor photoanode and photocathode in a single cell, enabling water splitting without an external voltage.

Organic semiconductor photoelectrodes offer broad light absorption and solution processability, but their performance can decline over time. Her research focuses on improving conversion efficiency and extending cell lifetime, with the longer-term goal of advancing solar hydrogen technologies for carbon-neutral energy systems.

“Receiving this scholarship is especially meaningful because it recognizes both my research and its potential for further development,” Jeon said. “I am grateful to Professor Han Hee Cho for his guidance and to my colleagues for their support throughout the research.”

JaeHun Bang, a master's student in the Graduate School of Artificial Intelligence, develops AI that can interpret visual context and connect natural-language requests with objects and events in its surroundings.

His research on video retrieval addresses a limitation of existing models, which often focus on prominent subjects while missing background details and changes over time. By segmenting videos at scene transitions and incorporating descriptions of both key subjects and surrounding context, he identified where current models struggle with complex contextual and temporal questions. The resulting paper, “Beyond the Highlights,” was accepted at WACV 2026.

He has since extended this work to 3D environments. LightSplat , accepted at CVPR 2026, allows users to locate objects using natural-language descriptions such as white sofa or egg on ramen, without relying on predefined object categories. The method requires about five seconds of preprocessing and, depending on the comparison, runs 50 to 400 times faster while using one sixty-fourth of the memory. Potential applications range from robots responding to spoken instructions to natural-language interaction in AR and VR and equipment or hazard searches in digital twins.

He previously conducted research on 3D scene understanding at Carnegie Mellon University and is currently an intern at LG AI Research, where he develops vision-language models for EXAONE. His goal is to combine fine-grained video understanding with language-based 3D perception to develop AI that can better interpret context and human intent.

“This scholarship gives me greater momentum to build on my research,” Bang said. “I am grateful to Professor Kyungdon Joo for his guidance, and I hope to develop AI that can communicate and collaborate naturally with people in unfamiliar environments.”

This year's Presidential Science Scholarship drew approximately 22 applicants for each available award. Selected master's students receive KRW 1.5 million per month, while doctoral students receive KRW 2 million per month.