DGIST Startup Univa to Lead 4.6 Billion Won National Research Project... Building Drug Development Data Through 2030

DGIST (President Lee Geonwoo) announced on May 14 that Univa Co., Ltd. (CEO Nam Myungjin), an AI-specialized startup founded at DGIST, has been finally selected as the lead institution for a subproject in the "AI x Bio Innovation Research Hub Pilot Project" promoted by the Ministry of Science and ICT, and will carry out a national research project worth 4.6 billion won.

The "AI x Bio Innovation Research Hub Pilot Project" is a large-scale national initiative aimed at leading the digital bio transformation by strengthening AI-based drug development capabilities. A total of 49.1 billion won will be invested, with government funding from the Ministry of Science and ICT (National Research Foundation of Korea) and matching funds from Daegu Metropolitan City, Kyungpook National University, and others. Kyungpook National University leads the consortium, which includes Daegu City, Univa Co., Ltd., Kyungpook National University Hospital, and the Daegu-Gyeongbuk Medical Innovation Foundation.

Nam Myungjin, CEO of Univa

Through this selection, Univa Co., Ltd. will execute a national research subproject worth a total of 4.6 billion won over five years, until December 2030. Univa will utilize the latest high-performance infrastructure (48 NVIDIA B300 GPUs) to build drug development datasets and will be responsible for developing an "AI foundation model" (a general-purpose AI trained on vast datasets for versatile applications) specialized for research domains.

Univa Co., Ltd., selected as the lead institution for the subproject, is an AI deep-tech company established in 2021 by undergraduate students from DGIST. The company possesses technology for building domain-specific datasets and proprietary foundation model source technology. Univa has demonstrated its technological prowess by building and publicly releasing vast, complex datasets—such as 520,000 scientific and technical documents and 600,000 mathematics and textbook data—through platforms like AI Hub.

Notably, in 2023, Univa’s Korean language-specialized model "KOMODO" ranked first on the open-source language model leaderboard (Open Ko-LLM). Recently, the company has also achieved continuous results, such as jointly releasing the inference model "DeepSeek-Bllossom" series with KAIST. Additionally, Univa has rapidly grown as a promising AI powerhouse, winning first place (Minister’s Award) at the Ministry of SMEs and Startups’ AI Championship and the DNA Innovation Award (President’s Award) from the National IT Industry Promotion Agency.

Going forward, Univa Co., Ltd. plans to use its AI foundation model to precisely predict drug candidate substances, immediately synthesize them in automated laboratories, and establish a "closed-loop" system—a continuously cycling automated system that connects predictions to validation in real time with Kyungpook National University’s activity evaluation data. Through this, Univa aims to safely utilize sensitive medical data within Daegu’s Data Regulation Free Zone and realize a seamless end-to-end AI ecosystem, actively targeting the global AI drug discovery market.

Nam Myungjin, CEO of Univa Co., Ltd., stated, "This project is a crucial opportunity for Univa to prove its expertise in building complex datasets and developing AI models," adding, "By combining top-tier clinical data and robust infrastructure, we will play a pivotal role in helping South Korea secure leadership in the global AI drug discovery market."

This content was produced with the assistance of AI translation services.