Rob Read

My personal dev projects, all work and ideas my own. Not all successful, learning every day.

Ground floor
Not to scale
Room 01

Living Home Survey

A plant-care mobile app adapted to the user's own environment: beyond identification and basic care, it works out where to keep each plant so it stays healthy. Right now I'm gathering feedback through a chat-style survey, where an OpenAI model interviews each respondent topic by topic and saves their answers to Supabase as they go. The survey app is self-hosted, deployed with Coolify.

  • Fixtures
  • Next.js
  • OpenAI
  • Coolify
  • DuckDB
Work in progress, planning towards 2027 release
Room 02

Golf AI

Experiments in AI swing analysis. A Python script reads a phone video of my swing with OpenCV, tracks 33 body points per frame with MediaPipe's pose model, then finds the address, top and impact positions to measure tempo. It draws a skeleton overlay and NumPy/Matplotlib diagnostic plots to test whether casual phone footage is good enough to build on.

  • Fixtures
  • Python
  • MediaPipe
  • OpenCV
Mini project · AI
Room 03

Trading App

Testing automated trading driven by signals, on a Trading 212 demo account. The bot runs as a service in my OpenClaw workspace, watches prices from Yahoo Finance and buys on RSI signals (mean reversion and short-term pullbacks, filtered by the 20-day average and overall market trend). Stop-loss, trailing-stop and take-profit rules handle exits, with each trade posted to Discord and a Next.js dashboard.

  • Fixtures
  • OpenClaw
  • Trading 212 API
  • Yahoo Finance
  • FastAPI
Active · demo account only
Room 04

Fitness & Health Coach

Helping me keep up a healthier lifestyle, with focused weekly coaching goals across weight training and cycling. It keeps an eye on my data so insights are easier to spot. Built simply within Claude, with a Supabase backend for my data and the Google Health API pulling in my watch data.

  • Fixtures
  • Claude
  • Supabase
  • Google Health API