Data Science Lead
- Designed, built, tested, and implemented custom predictive and prescriptive machine learning models to solve manufacturing, operational, and scrap challenges-- Programmed a prescriptive embedded neural network that diagnosis and corrects a failed machine cycle in real-time, saving ~$150k/year in scrap-- Applied random forests to extract meaningful factors affecting a 20-year scrap issue, saving $300k-- Manipulated large timeseries signals and used statistics and visualization tools to identify root cause of a 60% scrap issue, saving $200k in scrap-- Designed neural networks to predict properties of fired ceramic cores based on raw material properties, with >90% of test set performance within desired property range- Worked with engineering and operational experts to scope, build, and maintain Plotly visualizations and Power BI dashboards-- Dashboards have saved 100s of hours per year and identified root cause of multiple scrap issues, saving thousands-- Business-level dashboards allowed upper-level directors to gain actionable insights on key business metrics- Built and maintained large data ingestion pipelines from many sources (SQL data warehouses, Excel, Access, etc)- Ensured data quality by identifying and resolving issues - Optimized code for performance and documented code improvements using git version control- Presented technical results to director and executive-level leadership- Mentored interns and new employees
