Aeromag is a complete drone-borne magnetic survey system built to map buried magnetite over exploration ground in western Colorado: sensors, airframe hardware, flight configuration, and the data pipeline that turns a morning's flight into a ranked list of drill targets.
The Hardware Problem
A fluxgate magnetometer measures the Earth's field to fractions of a nanotesla — and a drone is, magnetically, a screaming nuisance. The system's governing constraint is magnetic hygiene: paired fluxgate sensors flown as a gradiometer (so the aircraft's own signature subtracts out), mounted on carbon tube with printed PETG fittings, nylon bolts, and not a ferrous screw anywhere. The entire mounting ecosystem — clamps, sensor pods, the towed "magbird" — was designed through the AI agent workflow documented in The Agent in the Workshop and printed in-house.
The Software Problem
Raw airborne magnetic data is unusable without serious geophysics, and the processing pipeline implements the discipline's canon in modern Python: ingest and QC of the 200Hz sensor stream, Tolles-Lawson compensation for aircraft motion, diurnal correction against USGS observatory data, minimum-curvature gridding, reduction-to-pole and analytic-signal derivatives, and Euler deconvolution tuned for compact near-surface sources — ending in anomaly picks ranked for follow-up. Each stage is a tested module; the orchestrator runs a flight from CSV to GeoTIFF without hand-holding.
Why It Matters Here
The system serves the West Side Placer program, mapping magnetic minerals across terraces too large to sample everywhere. One person carried the problem from what alloy can sit near the sensor to which regularization the gridder needs, using AI agents at every layer where they help.
