• Design an integrated system of wafer flow forecast and release policy suggestions to support supervisors in managing wafer release policies.• Enable fab throughput trend forecast for up to two weeks.

• Designed a pattern analysis algorithm and multiple indicators to quantify opportunities for enhancing wafer logistics efficiency.• Identified 10% potential improvement by eliminating inefficient manual interventions in wafer logistics.• Prioritized and targeted the most beneficial place for improving wafer logistics efficiency.

• Designed an autoencoder-based temporal and spectral feature anomaly detection system that identifies equipment anomaly during wafer processing without relying on Fault Detection and Classification (FDC) data labels.• Achieved 100\% detection accuracy on a dataset of 20 wafers.• Gained up to 8 days' earlier detection compared to the current control wafer monitoring method.

• Analyzed key factors of Fault Detection and Classification (FDC) data to balance monitoring costs and prediction accuracy of wafer defects using machine learning algorithms.• Selected 20 important factors out of 300 and achieved 75% prediction accuracy on the number of wafer defects.