BRKMYR/Prototypes I built
Independent product prototypes I designed and built with AI coding agents, separate from the commercial products I led (see Products I led). Three platforms, each worked as a product: a user with a costly problem, a vision, and the metrics that decide whether it works. One principle throughout: AI proposes, a deterministic layer explains why or says no. Where the platforms sit in the stack →
Platforms
Platform · Map data · Radar imagery
Ask maps and satellite radar a plain question. Get an answer with its evidence, on your own hardware.
North starVerified answers delivered without a specialist
Reference appsKartograf · SAR Analyst Console · Zurich 3D (Beta)
Open platform →
Platform · Data · Sim · Eval · Decide
The validation loop for physical AI. Simulate, score, decide, with every release traceable to a threshold set in advance.
North starTime from new model build to a defensible release decision
Reference appsLabel Quality Engine · World Model Benchmark · AV Safety Benchmark · Assurance Gate
Open platform →
Platform · Fleet · Multi-machine
One person supervises many machines. Software says which needs a human now and why, with every decision on record.
North starMachines per operator at constant safety
Reference appsFleet Safety Monitor · INTENT Operator Console
Open platform →
AI ResearchiCub Deep Learning Object Recognition · Master's thesis · a humanoid names objects it never saw
Components · each one a shipped reference appPlays on its own · hover to pause
Release candidate
NO GO