Weather · Forecast experiments
Atmos Local
A weather dashboard that brings backyard observations, ensemble forecasts, radar, and model comparisons into one place. Exploring how forecasts hold up against the weather that actually arrives.
On the workbench
Products, small experiments, and ideas taking shape. A place to share the work as it develops.
Marketplace · AI collaboration
A marketplace for interesting cars, boats, machinery, and other finds. Exploring how people and their own AI agents can research and work through a deal together.
Decisions · Shared work
A shared workspace for people and their agents to gather evidence, challenge ideas, and work toward a decision. An experiment in making collaboration visible.
Writing & publishing
A personal home for writing about technology and business, collecting useful links, and documenting what I’m building.
Weather · Forecast experiments
A weather dashboard that brings backyard observations, ensemble forecasts, radar, and model comparisons into one place. Exploring how forecasts hold up against the weather that actually arrives.
Robotics · Simulation
A mobile robot experiment in Webots, exploring how an agent can navigate rooms, inspect objects, and attempt small manipulation tasks. Camera evidence and physical scoring are tracked separately.
Making · Parametric design
A growing library of small objects and mechanisms built with OpenSCAD and a Bambu printer. Keeping the design, sliced build, and observed print result together makes each iteration easier to inspect.
Agents · Home hardware
A local toolkit for reading a Roomba’s status and giving an agent a small set of cleaning controls. Exploring useful device access while keeping credentials and unrestricted commands out of the tool interface.
Email · Classification
An email review tool that separates category, reply intent, action, and urgency. It keeps model predictions and human corrections side by side so uncertain messages can get a closer look.
Voice · Practice & feedback
A prototype for turning customer-call transcripts into scorecards and focused practice. Exploring whether specific feedback and AI roleplay can give teams something useful to work on between calls.