I build demand-forecasting systems used in planning and S&OP — from research and temporal validation through APIs, databases and analytical interfaces.
At nPLAN (NEO Digital Industries) I developed 20+ forecasting models across statistical methods, machine learning, deep learning, foundation models and ensembles. That work raised accuracy by about 15% over models already used by large companies, and cut processing time from roughly 8 hours to 2.
Python · C# / .NET · PostgreSQL · React / TypeScript
Backtesting · Multiple metrics · Exogenous data · GPU / CUDA
Selected work: theylor.vercel.app