Proof capsule
Argentina Poverty Measurement & Atlas
End-to-end data engineering, validation, modeling, and public data product
A versioned survey-to-census data system with semantic alignment, predictive modeling, commissioning, and a public geographic atlas.
Problem
Survey and census microdata have different coverage and semantics. Publishing geographically detailed estimates requires explicit alignment, model-support checks, reproducible releases, and clear distinctions between observed and predictive results.
What I built
A multi-repository system connecting EPH household surveys, census microdata, semantic alignment, person-level income modeling, commissioning checks, governed releases, and a public atlas with geographic exploration.
What this proves
- Build governed data pipelines across heterogeneous public datasets.
- Distinguish observed, fitted and predictive estimates and preserve geographic support boundaries.
- Ship inspectable analytical releases consumed by a public web application.