Wai Key, Lee

Team lead of data science operation and compliance, driving outcomes in the following projects:

• Predictive Underwriting: Enhanced insurance underwriting model to reduce underwriting slip through and increase straight-through processing (STP) rates.
• Medical Claims Anomaly Detection: Improved anomaly detection model, with a primary focus on fraud, wastage, and abuse detection, to reduce erroneous medical claims payouts.
• Medical Bill Processing: Conducting proof of concept for image classification and OCR-based itemised charge extraction from medical bills, achieved up to 60% cost savings.
• Claim Fraud Detection Engine: Developing SAS fraud detection engine for non-medical claims, leveraging machine learning on top of life claim rules and social network analysis.

• MLOps research & implementation: Version Control (DVC), Unit Testing (PyTest), Experiment Tracking (MLFlow), Model Deployment (BentoML), Model Monitoring (Arize), Infrastructure Management (SageMaker)
• Developed AEON credit scoring model for AI showcase in digital banking license application
• Implemented Continuous Training (CT) pipeline for loan underwriting model using various AWS services
• Conducted POC of account level and transfer level fraud detection models
• Developed and deployed Model Cards application using Streamlit

• Developed and deployed interactive reporting dashboard using Plotly Dash (Python)
• Designed and Implemented Python-based ETL pipelines, connecting various data sources (e.g., MSSQL, PostgreSQL) and automating with Airflow
• Conducted POC of Screw Conveyors Failure Anomaly Detection using Sequence Models.
• Developed Speed Process Control for Lost Cook Equipment Preventive Failure
• Applied ML algorithms for production chemicals optimization