• Improved inspection efficiency and accuracy by using VGG-19 and RF model (With large data and fine labeling, the accuracy improved 34% when compared with the based classifier) • Fine-tuned every model such as data preprocessing and hyper-parameter tunning • Compared different CNN models and finalized the best model for inspection systems in plants’ line • Programmed scripts using Python for further data analysis (improving models)

• Improved accuracy from 65% to 97% for Pupil tracking system when finding Region of Interest(ROI) by implementing YOLO object detecting system using Python. • Programmed scripts using Python for labeling training data • Wrote manual scripts to teach other team members how to use the model for finding ROIs in the Pupil tracking system

• Improved inspection efficiency and accuracy by programming a script for operators using Halcon and C (Can detect 4 times more defect images than the original program) • Developed rules for classifying defect images by using image processing techniques • Secured Best Presentation format in the final presentation contest • Participated in Six Sigma training programs and learned concepts such as DMAIC and DMADV

Responsible for solution support for the Connected Vessels Project of U-Ming Marine. It was a global project which provides both data and voice communication between vessels and fleet management at U-Ming HQ