DGIST Student Startup UNIVA Co., Ltd. Selected for Hyperscale AI Ecosystem Expansion Project

- Carrying out two projects, including “Chain-of-Thought (CoT)-Based Manufacturing and Logistics Robot Data” - Developing CoT-based training datasets for industrial applications

□ UNIVA Co., Ltd., a student startup from DGIST (President Kunwoo Lee), led by CEO Myung-jin Nam, a graduate of DGIST’s School of Undergraduate Studies, has been selected as the lead organization for two projects under the 2025 Hyperscale AI Ecosystem Expansion Project. This initiative isorganized by the Ministry of Science and ICT and the National Information Society Agency (NIA).

 

□ Starting July 1, UNIVA will spearhead two projects: ▲”Chain-of-Thought (CoT)-Based Manufacturing and Logistics Robot Data” and ▲”Molecular Toxicity Inference Data.” The “CoT-Based Manufacturing and Logistics Robot Data” project is a large-scale initiative with a total budget of KRW 1.37 billion. This project aims to apply CoT, a core feature of hyperscale language models, to the manufacturing and logistics industries. The focus will be on developing high-quality training datasets that enable collaborative and logistics robots to autonomously manage the entire process of perception, decision-making, and execution.

 

□ This project is being conducted under UNIVA’s leadership in collaboration with partner organizations such as Gunn Solution, the Gwangju Institute of Science and Technology (GIST), Seoul National University of Science and Technology, and C&B System. The official project kickoff took place in July, marking the start of full-scale research activities.

 

□ Throughout this project, UNIVA plans to accelerate the adoption of autonomous robots in the manufacturing and logistics industries by leveraging the CoT-based datasets it will develop. It also aims to create a flagship use case to validate the applicability of language and vision models in industrial settings. In addition, by standardizing URDF and CoT-based data, UNIVA seeks to ensure compatibility and reusability across various robot platforms, fostering a data-driven ecosystem within the domestic robot AI industry.

 

□ In recognition of its technological capabilities and research achievements both in Korea and abroad, UNIVA was recently awarded the “Minister of SMEs and Startups Award” in the AI Excellence Award for Enterprises category at the AI Excellence Summit Awards, held as part of the 2025 AI EXPO KOREA.

 

□ Founded as a student startup at DGIST, UNIVA develops and provides industry-specific AI technologies based on AI training data and language model solutions. More recently, the company released its Korean-specialized AI model series, DeepSeek-Blossom, as open source, gaining significant attention in both domestic and global AI ecosystems.

 

□ Meanwhile, DGIST offers various programs, including the DGIST-TIPS (Technology Incubation Program for Startups) and the Open Innovation Program, to actively support technology-based startups by faculty, researchers, and students. Through these initiatives, startups are strengthening their technological capabilities and expanding opportunities for commercialization, while those with outstanding technologies are accelerating efforts to establish a foundation for global expansion, enabling their entry into international markets.