**AI testing platform Development and Maintenance**- Designed workflow and relational database according to normalization and use scenarios.- Built testing engine which integrates any model or API of PyTorch, ONNX and OpenCV.- Created a website with a backend engineer for visualization.- Cross-platform deployment by using docker.- Used by NARLabs for testing AI model robustness.**AI security system Development and Maintenance**- Defined use cases with business teams and designed a multi-threaded behavior detection algorithm which can detect specific behaviors accurately in real-time.- Managed the backup images of unusual behaviors on AWS S3.- Solved the problem of network and incompatible packages on edge devices.- Used by Taiwan Taoyuan International Airport for monitoring confidential area.**Wi-Fi Data Pipeline Design and Build**- Designed relational database according to normalization and use scenarios.- Built ETL pipelines from Elasticsearch to MySQL to offer personal features, which could classify people into different types of customers.- Collaborated with business teams to design data visualization for the dashboard.- Conducted data analysis reports to illustrate the results and insight.- Used by 30+ places across the nation including tourist spots such as Taipei Zoo.**ML Model Lifecycle (Side Project)- Built Image Classification pipeline by writing efficient vectorized code with numpy and PyTorch.- Built KNN, Softmax classifier, SVM classifier and Neural Networks.- Implement various training technique such as batch normalization, regularization, vectorized backpropagation and update rules.- Understand the differences and tradeoffs between image classifiers.