Projects
Ministry of Science and ICT (MSIT), Korea · 2025–Present · Participant Researcher
InnoCORE Project
National flagship program (~$25M USD over 4.5 years) to advance AI-driven manufacturing and foster postdoctoral researchers across Korea's four Institutes of Science and Technology (KAIST, GIST, DGIST, UNIST). Led proposal development and conceptual design of an LLM-based autonomous manufacturing framework integrating materials, structure, and processing.
KIRD · Apr–Oct 2026 · Principal Investigator
CLAMP — Postdoctoral Learning Community (CoP)
Postdoctoral learning community (~₩8M support) conducting interdisciplinary study on knowledge-informed AI methodologies, integrating physics and domain knowledge into AI models for materials design, structural analysis, and process optimization.
NIPA · 2026 · Participant Researcher
Advanced GPU Utilization Support Program
Competitively selected for NIPA's Advanced GPU program under AiM4 Lab, KAIST. Research on Continual Pre-training and industrial application for building design & manufacturing domain-specialized foundation models. Allocated NVIDIA B200 × 7 servers (56 GPUs) for 2 months. Led proposal writing for the project.
Korea Electronics Technology Institute (KETI) · 2026–Present · Participant Researcher
Large Action Model-Based Digital Twin Autonomous Operation for Active Problem Solving
Commissioned research (~₩120M/year) under AiM4 Lab, KAIST, developing Large Action Model (LAM)-based digital twin technology for autonomous operation and active problem solving in intelligent manufacturing systems. Led proposal writing for the commissioned portion of the project.
Ministry of Science and ICT (MSIT), Korea · 2025–Present · Participant Researcher
National Agenda Basic Research Program
Three-year national research project (~$150K USD/year) focused on developing domain-specialized LLMs to enhance productivity in advanced manufacturing. Contributing to core LLM frameworks and application technologies enabling intelligent, domain-aware automation in smart manufacturing workflows.