ParkingMate.ai
Making parking rules easier to understand, one sign at a time.
A better place to park.Vision + local context
A small everyday problem.
A surprisingly complex system.
Parking signs pack conditions, time windows and exceptions into a very small space. ParkingMate.ai combines sign interpretation with local-context retrieval to turn those details into structured decisions and plain-language guidance.
From model output to a usable product.
I built the application across its interface, AI integration and backend, including authentication and Stripe payments. The work spans prompt refinement, retrieval, tool-use workflows, testing and deployment.
Sign image → AI interpretation → Local context → Structured guidance
Learning from the edges.
The hard part is handling ambiguous signs and incorrect interpretations responsibly. Real-world failure analysis and user feedback inform how I refine prompts, output formats and application behaviour.
What this project brings together.
Applied AI, end-to-end product development, and the less glamorous but essential work of making a useful application: clear interfaces, payment flows, testing and continuous iteration.