Skip to content

Featured Projects

Welcome to my portfolio of geospatial data science and AI projects. Each project demonstrates my expertise in delivering impactful solutions to real-world business challenges.

  • End-to-End Yield Monitor Data Processing Pipeline

    Automated workflow that transforms noisy, unreliable yield monitor exports into clean, enriched, and spatially consistent analytics layers. The pipeline processes 130,000+ hectares per season, applying multi-header calibration correction, edge point removal, spatial enrichment, and H3 hexagonal indexing for cross-campaign comparison in large-scale farming operations.

  • Water Risk Assessment for Agricultural Land Acquisition

    Satellite-based historical water risk analysis using Google Earth Engine and Dynamic World to characterize flood probability across agricultural fields. The workflow processes 6+ years of Sentinel-2 imagery, producing monthly risk profiles and per-parcel comparisons that directly support farmland purchase and leasing decisions.

  • Hydrological Mapping: Automated Waterlogging Detection at Field Scale

    Automated geospatial pipeline using FABDEM, Google Earth Engine, and WhiteboxTools to identify topographic depressions and drainage pathways across agricultural fields. Delivers 50 m buffer restriction zones for silo bag placement and frost-risk assessment — processing 130,000+ hectares per run with zero manual digitizing.

  • Beyond the Field Visit: Automating Crop Performance Analysis for Farmland Lease Decisions

    Automated remote sensing workflow to benchmark NDVI performance of a candidate field against neighboring lots growing the same crop in the same season. Using Sentinel-2 imagery and the Google Earth Engine Python API, the system produces interpretable histograms that contextualize the field within a 20 km radius agronomic environment to support farmland lease decisions.

  • Automatic phytosanitary application compliance mapping

    Automated GIS system for calculating phytosanitary application restriction zones across Argentine agricultural fields, processing multi-jurisdictional regulations to generate compliance maps and reports for aerial and terrestrial spraying operations.

  • Semi-Automatic Land Cover Classification with Sentinel-2

    Semi-automatic land cover classification of a 12,756-hectare cattle ranch in San Luis, Argentina, using Sentinel-2 imagery and unsupervised machine learning. Ten distinct land-cover classes were discriminated in under one week — a 75% time reduction compared to manual mapping — delivering actionable maps and area statistics for agricultural conversion planning.

  • Large-Scale Distance Matrices with Local OSRM Server

    Running OSRM locally enables fast, unlimited calculation of large origin–destination distance matrices using real road networks. This approach removes public API limits and helps companies model logistics costs accurately by computing kilometers between thousands of origins and destinations.

  • GeoNode Platform Deployment for Professional GIS Presentation

    Implementation of a GeoNode-based web geospatial platform to publish and share professional cartographic projects. Deploying on a dedicated server using Docker Compose resolved technical port conflicts and health check issues, enabling full control and customization of how GIS work is presented for clients and organizations, with secure access and high availability.