Customer Credit Evaluation System
• Developed a risk management model using XGBoost to assign customer credit ratings for credit term and limit decisions, managing $20B+ in customer value, reducing overdue repayment cases by 14%, and improving auditing efficiency by 95%.
• Fine-tuned XGBoost with Optuna for hyperparameter optimization, utilizing feature engineering, SMOTE oversampling for class balancing, and WOE (weight-of-evidence) encoding to address data quality issues and improve accuracy.
• Deployed the ML pipeline on AWS SageMaker with automated workflows, runtime configurations, and database integration.
Automated Inventory and Production Workflow Optimization System
• Engineered a Python-based Robotic Process Automation to automate manual procedures in inventory allocation and production planning for a glass manufacturing company, reducing annual delayed shipment by 30%, and increasing process efficiency by 95%.
• Developed APIs to integrate the RPA script with the client's internal platform, enabling event-driven execution of cloud-based workflows and real-time display of model results on the front-end interface.
