I investigate, build, and learn.

I follow questions through software, systems, hardware, and observation. These selected records offer a few ways into the work: what I tried, what the evidence showed, and where the limits remain.

All work & notes

Find the failing boundary

A silent board, a camera protocol, or a compatibility layer can hide several different problems. These investigations separate the observation from the assumption and record what actually worked.

Test models under constraints

Memory budgets, interrupted training sessions, and the gap between simulation and hardware make useful limits concrete. The outcomes here are specific to the devices and tests described.

Make information visible

Maps and diagrams offer another way to investigate a system. These records cover map-loading work and recovered visual studies; the archived studies are not presented as live interactive tools.

Look closely, keep a record

Not every question starts with a technical fault. A photograph, a family memory, or an old object can be worth following—and worth recording even when the result is uncertain or the idea is set aside.

All work & notes