Agricultural researchers needed to run complex biogeochemical simulations but existing tools required desktop installations, manual data formatting, and deep technical knowledge of model internals.

Built a 4-tier microservices platform connecting an interactive mapping frontend to model orchestration services. Users draw field boundaries, enter management practices, and receive emissions projections — the system handles model translation, execution, and results visualization. Supports DayCent and Landscape DNDC engines with a plugin architecture for additional models.

A global building technology company needed to extract structured data from 50+ page vendor inspection PDFs — a manual process taking 8 hours per report across multiple report formats and vendor-specific terminology.

Deployed an on-premise Intelligence Node with multi-path document extraction (rule-based parsing for 99.8% of PDFs, Claude vision API fallback for scanned documents). Built a three-tier classification engine with confidence scoring and a human-in-the-loop learning loop — every technician correction automatically improves the knowledge base for future processing.

Conservation scientists needed a way for landowners to assess soil organic carbon and water capacity at parcel level, with decision support for compost application planning across California counties at 30m resolution.

Built a hybrid rendering geospatial application combining client-side vector tiles (PMTiles via MapLibre GL) with server-side raster WMS (GeoServer). Node.js backend orchestrates Python GDAL processes for raster operations. Includes point-click analysis, area-of-interest drawing, zonal statistics, and 1-30 year benefit projections.

A 3,200+ home smart community needed a unified platform to manage the complete lifecycle from county lot filing through 2-year post-closing homeowner support — coordinating builders, technicians, and 15,000+ IoT devices.

Built a hybrid enterprise architecture combining an ASP.NET Core web platform with a Python CLI filing pipeline. The system manages 4-phase home commissioning, automated lot filing (500+ homes processed in under 30 seconds), drag-and-drop technician scheduling, and real-time IoT utility monitoring across gas, electric, and water meters.