Strategy Development, focusing on Taiwan Stocks and ETFs.

• Data-driven Decision Making — Coded for user activity data, such as DAU, WAU, Average Active Day, and Cohort Analysis, etc.— Used funnel analysis to find that only 35% active users experienced our search engine in one new version because of the complicated setting, and finally raised it to 63%• Cashback Data Format Building and Maintained Cashback Data— Quantified complicated cashback description iteratively so as to display correct information simply— Built checking… Show more • Data-driven Decision Making — Coded for user activity data, such as DAU, WAU, Average Active Day, and Cohort Analysis, etc.— Used funnel analysis to find that only 35% active users experienced our search engine in one new version because of the complicated setting, and finally raised it to 63%• Cashback Data Format Building and Maintained Cashback Data— Quantified complicated cashback description iteratively so as to display correct information simply— Built checking pipeline for part-time workers to ensure cashback data quality• User Interview— Ensured user recognition of Cardbo, knew how they use and what they liked Show less

Research Field: Computer Vision with Deep Learning MethodTopic: Image Inpainting and Extrapolation— Restored missing parts of corrupted images in deep learning-based methods— Used adversarial learning methods to make the results more realistic

Research Field: Computer Vision with Deep Learning MethodTopic: Multi-Source Domain Adaptation in Image Classification— Improved classification accuracy when the image style of train data was different to the one of testing data — Won the first place in class & the second place in 2019 NTUEE Undergraduate Innovation Award