Parkly

Inspiration

Sydney drivers waste 15+ minutes circling for parking. NSW Government data shows Liverpool CBD hits 85% midday occupancy. No single app tells drivers where spots actually are.

What it does

Pick a destination → Parkly ranks every nearby spot by availability, commute, walk, and price. Traffic-light status with a three-tier confidence model: Tracked (95%, TfNSW live API), Derived (75%, Park'nPay tickets), Predicted (50%, heuristic). Users sign in and reserve bays through a real backend.

How we built it

Three processes: parkly.html (Leaflet map + ranked list), proxy.py (CORS relay for TfNSW), backend.py (Flask + SQLite, auth + bookings). Three real NSW datasets embedded: 23 Park&Rides (live), 76 CBD commercial lots, Loading Zones. Scoring: availability×35% + commute×20% + walk×20% + price×15% + confidence bonus − busy penalty.

Challenges

CORS blocked direct API calls — solved with local proxy.

What we learned

Being explicit about data confidence is itself a feature.

What's next

Park'nPay feed · hunt-time prediction · Wilson/Secure partnerships · Melbourne expansion.

Built With

• city-of-sydney-open-data
• css
• flask
• html
• javascript
• leaflet.js
• openstreetmap
• python
• sqlite
• tfnsw-api