Rodolpho Justino

At AkzoNobel, I lead end-to-end demand forecasting solutions for portfolio forecasts across business units in the Americas and Europe. I build models in Azure Databricks using Python, PySpark, SQL, machine learning, and time-series methods.

I develop forecasting approaches with XGBoost, StatsForecast, Prophet, ARIMA, and ETS, using time-series classification and segmentation to select models by product behavior. I also implement backtesting and comparison frameworks using WMAPE, MAPE, and RMSE.

Previously, I delivered Power BI and ETL-based business intelligence solutions for mid-sized companies, structuring operational and commercial KPIs for faster decision-making.

My research with UFCG and Petrobras combined mathematical modeling, supervised machine learning, gas-dispersion simulations, and statistical analysis for hazardous industrial-area classification.