/w=1920,quality=90,fit=scale-down){kind=small}
I build small, fast experiments to explore ideas outside my day job - tools, prototypes, and community projects that help me think, test assumptions, and turn messy questions into something tangible. Some are serious, some are playful, but they all teach me something about design, AI, or how people actually use things.
.svg)
Streamsight
A proof of concept exploring how to turn the many insights we generate at work into a searchable RAG database. Built with Google Gemini, custom indexing, structured notes, and user authentication.
What I learned:
How to build usable RAG workflows, design search-first tools, and bring order to large pools of unstructured insights.
Related writing:
East Village Hub
A community portal for East Village, London - with a directory of local businesses, area information, groups, societies, and recommendations from residents.
What I learned:
How to design hyperlocal structures, onboard non-technical contributors, and build trust in community-led digital spaces.
Dynasty League Roaster
An AI-powered app connected to the Sleeper API, pulling weekly results and generating personalised roast messages - often crossing the line from fun into pure mischief.
What I learned:
API integration, personality-driven AI generation, and quick prototyping that deliberately sits on the edge of chaos.
.svg)
Secret Santa
A lightweight, privacy-first Secret Santa that never stores your data. Names go in, pairings come out, and everything else vanishes. Vibe-coded. No accounts. No database. Just ephemeral pairing. Built fast to prove you can make festive tools without harvesting anyone’s info.
What I learned:
- Privacy-by-design patterns with zero data retention
- Client-side generation and ephemeral state
- Keeping “just enough UI” for non-technical groups