Wei Zhang
Senior Data Scientist | Pricing, Churn and Forecasting Models in Production
Summary
Senior Data Scientist with 7 years building predictive models for insurance and energy companies. Takes projects from framing the business question to deployed models with monitoring. Strong in Python, SQL, gradient boosting and experiment design, and known for explaining model trade-offs clearly to actuaries, product leads and regulators.
Experience
Senior Data ScientistApr 2022 - Present
Harborline Insurance
- Built a renewal churn model (LightGBM, AUC 0.81) that targets retention offers; lapse rate among targeted policyholders fell 2.4 points, worth about $3.1M in retained premium a year.
- Designed the A/B test framework the pricing team now uses for all rate changes, with 14 experiments run since 2023.
- Worked with actuaries to add telematics features to the auto pricing model, improving loss ratio prediction by 9% on holdout data.
- Moved 6 models from notebooks to scheduled pipelines on Databricks with drift monitoring.
- Mentor 2 junior data scientists.
Data ScientistAug 2019 - Mar 2022
Cascade Energy Analytics
- Forecast hourly electricity demand for 4 utility clients, with mean absolute percentage error of 3.2% day-ahead.
- Built an anomaly detection model on smart-meter data that flagged 1,100 likely meter faults in its first year.
- Presented model results to utility regulators in 3 rate case filings.
Skills
- Python (pandas, scikit-learn, LightGBM)
- SQL
- Machine Learning
- Experiment Design and A/B Testing
- Time Series Forecasting
- Databricks and Spark
- MLflow
- Model Monitoring
- Causal Inference
- Data Storytelling
Education
MS in Statistics, University of Washington, 2019; BS in Mathematics, University of California, Davis, 2017