Things I built.

An applied research project that turns Garmin data, clinical literature, nutrition logs, and AI analysis into a more interpretable view of personal health. It is less a finished product than a working argument about what individualised health software could learn from evidence, context, and longitudinal data.

I automated as much of my role as I could, stepped into the fully-automated version of my function, and found out what remained. Six findings on decision architecture, schema annotation layers, expert intuition as an input channel, judgment sequencing, and the drawer of commercially dead problems that AI has brought back to life.