Builder at the intersection of AI, product thinking, and real-world systems.
My approach starts with observation — I look at how systems actually behave, identify friction that's easy to overlook, and work out what's worth solving. From there, the path to execution is short. I use Google Antigravity, Claude Code, and terminal-based workflows to move quickly from insight to working prototype, integrating open-source tools pragmatically along the way.
Most of what I build started as something I needed myself. That's where the ideas come from — lived experience, not briefs. My value is in the full loop: spotting what needs to exist, understanding why it matters, and building enough to prove the idea works — validation over perfection.
| Project | What it does | Stack |
|---|---|---|
| tote-counter | AI vision tool that counts warehouse tote stacks using Llama 4 via Groq — 69% exact accuracy via chain-of-thought prompting | Python · FastAPI · Groq |
| CantStopLearning | Android app that generates AI audio lessons on any topic via Perplexity Sonar API + TTS | React Native · TypeScript · Supabase |
| stridemind | Walking companion app — capture timestamped insights mid-walk via Health Connect | Flutter · Dart · SQLite |
| shoe-fit-biomechanics | Interactive research documentary on the engineering of shoe discomfort | HTML · CSS · Vite |
| Study | Focus |
|---|---|
| Amazon Delivery UX Proposal | UX + financial modelling — £168M ROI case for a single notification UI fix |
| Perplexity Voice Mode Study | Feature proposal for voice conversation continuity with impact analysis |
| Amazon Envelope Analysis | SQL-backed process improvement from firsthand warehouse observation |
Languages: Python · TypeScript · Dart · JavaScript · SQL
Frameworks: FastAPI · React Native · Flutter
Backend: Supabase · SQLite · Groq API · Perplexity Sonar API
Tools: VS Code · Claude Code · GitHub · Render