CoffeeMap.dev - Geospatial Engineering Platform

Back to Home

GeoAI & Remote Sensing

Machine learning and satellite analysis using Google Earth Engine for actionable intelligence.

Google Earth Engine

Python

D3.js

Chart.js

Pix4D

Projects

GeoAI - Hydrology

Blue Nile Flood Risk

Client-side GeoAI flood risk mapping for Sennar and Gezira, Sudan — Sentinel-1 SAR flood detection, DEM slope, rainfall and river distance combined in an adjustable risk model, all computed in the browser.

Blue Nile Flood Twin

An animated 3D flood of Sennar, Sudan during the September 2020 Blue Nile flood — terrain, extruded buildings, rising water, rain following rainfall and buildings coloured by water depth, with the same animation exported to video.

Riyadh Urban Heat & Tree Canopy Equity

Client-side GeoAI for Riyadh — Landsat surface temperature hotspots, Sentinel-2 NDVI and tree canopy, OSM trees and parks, a TensorFlow.js heat-risk model and canopy-equity priority zones for planting, all computed in the browser.

GeoAI Crop Monitoring

Sentinel-2 crop monitoring for irrigated farms in Al Jouf, Saudi Arabia — auto-detected centre pivots, NDVI/EVI time series, crop classification and anomaly alerts.

Serverless GeoAI Architecture

Interactive architecture of the GeoAI dashboards: two live routes (compute in the browser, or precompute once and stream imagery) and a planned third for trained models, with no VPS, database or Python at runtime.

Insights

GeoAI

28 Sept 2026

Serverless GeoAI: The Architecture Behind CoffeeMap.dev

How three satellite dashboards run with no VPS, no database and no Python at runtime: two routes, six patterns and one rule for where each computation belongs.

GeoAI

28 Sept 2026

Flood Risk Mapping on the Blue Nile, Entirely in the Browser

Sentinel-1 radar, terrain tiles, rainfall and OpenStreetMap rivers become a flood-risk screening model for Sennar and Gezira, with no server at all.

GeoAI

28 Sept 2026

Where Riyadh Needs Trees Most

Landsat heat, Sentinel-2 vegetation, WorldCover and OpenStreetMap on a 1.6-million-cell grid, client-side, to rank the blocks where one more tree matters most.

GeoAI

28 Sept 2026

Crop Monitoring for Al Jouf Without a Server

Centre-pivot detection, NDVI/EVI time series, crop classification and anomaly alerts, built offline in Python and served as static files plus live tiles.