Here are some of the projects I’ve worked on, each one focused on solving a specific problem with data and machine learning.
🔋 Battery Swapping Station Location Optimization
Geospatial modeling to support where to install battery stations across Korea.
Used population, rider mobility, and slope data to predict local demand.
Tools: XGBoost, geopandas, folium
🐶 Dog Facial Expression Synthesis with StarGAN_v2
Fine-tuned a GAN to generate emotion-based dog faces while keeping breed and fur color intact.
Focused on disentangling identity from expression.
Tools: StarGAN_v2, EfficientDet, histogram loss
📚 Policy Knowledge Graph from Korean Development Reports
Extracted entities and relations from Korean policy reports to build a searchable knowledge graph.
NER with BERT+CRF, relation extraction with R-BERT.
Tools: BERT, R-BERT, Neo4j
🛠 Anomaly Detection for Welding Defects
Built an anomaly detection system using sensor signal transformations and CNNs.
Used in real production to flag subtle welding failures.
Tools: STFT, Isolation Forest, 1D-CNN
🔥 Electrode Surface Temperature Forecasting
Deployed a custom time series model for surface temperature control during battery manufacturing.
Helped shift from rule-based to model-based control logic.
Tools: LSTM, Attention, factory deployment