Backend/2026/Production
RideSync+
Puts Uber, Bolt, Yango, and Shaxi side by side using real road distances from Google, forecasts surge with a RandomForest model, and streams live prices to a Flutter app over WebSockets with push alerts. It was my final-year project, and it ships with a full test suite.
Context
In Ghana there is no easy way to compare Uber, Bolt, Yango, and Shaxi at the same time, and surge pricing is basically a black box. RideSync+ brings them into one place and forecasts surge before it actually hits.
Approach
- 01Aggregated four platforms behind a platform-adapter pattern, pricing trips on real Google road distances rather than straight-line estimates.
- 02Ranked options with weighted multi-criteria decision analysis and forecast 24-hour surge with a RandomForest model that auto-retrains every six hours via Celery Beat.
- 03Streamed live prices over WebSockets and pushed alerts through Firebase Cloud Messaging, with Redis caching and Alembic migrations.
- 04Paired the FastAPI backend with a Flutter mobile client and a 49-test suite.
Notable
- BSc final-year project delivered to production quality.
- Self-maintaining ML pipeline with scheduled retraining.
- Real-time pricing and push alerts across four ride platforms.
Next project
Araba — University AI Assistant