At Yuanta Commercial Bank, I designed and productionized a fraud detection model that generated high-risk case lists across 149 branches and reduced false-alert volume by approximately 80%.
I also introduced Neo4j to build a financial knowledge graph and visualize transaction relationships, reducing case investigation and verification time from approximately 20 working days to around four.
At Cohesion Information Technology Ltd., I built medical-data clustering models that isolated chest X-rays with 99% precision for academic COVID-19 research. I also contributed to reimbursement auditing models and integrated AI features into an internal platform through Flask-based REST APIs.
Earlier, as a Data Scientist Intern at OmniHealth Group, Inc. (Taiwan), I developed ANN-based models for heart-rate variability analysis and prepared medical images for diagnostic AI. My work has also included production ML operations, medical imaging, and co-inventing two granted Taiwan utility model patents.

