Hugo Carpena

I’m a geospatial data scientist who turns messy, real-world raster sources into clean, validated vector geometries. I build ML-based vectorization pipelines from scanned and incomplete historical maps, then translate those outputs into usable spatial evidence.

My GIS workflow is end-to-end: I use QGIS and ArcGIS for spatial visualization and analysis, and I rely on geoprocessing tooling such as GDAL and spatial databases like PostGIS to support robust, repeatable spatial processing. I’ve also worked directly with geocoding and choropleth mapping, including territorial equity-focused analysis.

A key strength of mine is using spatial structure as evidence. For example, I’ve reconstructed historic street networks from extracted geometries and used measures of regularity and grid order as analytical instruments to test how urban form relates to air pollution outcomes across Spanish municipalities.

I’m equally rigorous about uncertainty and defensibility. I design and stress-test analytical pipelines before conclusions are drawn, and I apply causal inference methods for imperfect spatial and policy data, including Difference-in-Differences, Synthetic Control, Instrumental Variables, Regression Discontinuity, and Propensity Score Matching. I also communicate complex spatial relationships clearly for research and policy audiences.