Yihan Liu Email and Phone Number

◦ Led the development and deployment of graph models including LPA and GNN and clustering models to detect fraudster groups offline for loss estimation and ground truth for real-time models development.◦ Built end-to-end pipelines for model development, deployment, governance, and recovery actions.◦ Developed and deployed a suite of machine learning models (Random Forest, XGBoost, Transformers, etc), to identify and block suspicious transactions in real-time.◦ Collaborated with cross-functional teams, designed and developed new model feedback mechanisms and products, and onboarded new data sources and APIs to drive continuous model improvement.◦ Performed continuous monitoring and in-depth analysis of bad cases to identify emerging fraud patterns and proactively enhance model performance.

Developed machine learning models to improve banks’ transaction monitoring systems- Applied supervised machine learning classification models to flag out suspicious money laundering transactions- Used unsupervised machine learning models to cluster bank customers to increase the efficiency and effectiveness of the transaction monitoring algorithms- Built an anomaly detection model- Various text clustering models were implemented including LDA and CNN to determine the optimal model and approach to cluster vendors based on the descriptions of purchasing order documents- Identified anomalies using Local Outlier Factor algorithmAnalyzed transactional and customer profile data to identify suspicious transactions and potential fraud- Applied natural language processing techniques to clean unstructured data and identify namevariations and related parties- The funds flows between customers were visualized and analyzed using network graph- Potential fraud and money laundering activities were flagged out and highlighted to managementImplemented various data analysis methods to identify risks and frauds, including data massaging, data quality check, data exploration, identification of unusual transactions, mispricing and keywords searchBuilt interactive dashboards and reports using Tableau and Qliksense to visualize data analysis outputs

Inventory Analysis and management- Demand forecasting and price analysisDeveloped interactive web-based dashboards and reports by QlikView to visualize business insights and analyze business performance.