Stelios Papargyris

I’m a Data Scientist focused on the intersection of AI, weather, earth observation, and environmental data—turning complex geospatial and atmospheric signals into decisions. I’ve worked on production-grade pipelines, from ECMWF ensemble post-processing to real-time precipitation nowcasting across Europe.

At WeatherXM, I designed and deployed an XGBoost bias-correction system for ECMWF 51-member seasonal forecasts against ERA5, delivering +52% HSS improvement for precipitation and 30–50% RMSE reduction for temperature (plus 25–29% RMSE reductions across wind speed, humidity, and pressure). I also built an automated parametric insurance resolution system at ParahubXM using ERA5-based threshold precipitation analysis to classify risk and resolve contracts.

I’m hands-on with end-to-end ML reliability and testing: I contributed to Quality of Data/Quality of Forecasts pipelines (MLflow edge cases, QoD logic) and built a comprehensive automated pytest suite for forecast evaluation and data transformers. I also self-built Precipiteau, a 24/7 European rainfall nowcasting platform fusing radar, NWP, and satellite data with live products updated every 15 minutes.