01
FeedBook
- TYPE
- Android app
- ROLE
- Sole developer
- YEAR
- 2026
- Java
- Android
- Supabase
- PostgreSQL + PostGIS
- Retrofit
- Room
- osmdroid



THE PROBLEM
The same grocery item costs different amounts from store to store, and the government's suggested retail prices don't tell you where to buy.
WHAT I BUILT
A community price map. Shoppers log prices they see, the map compares stores within 2 km and against the suggested price, and a monthly shopping list is re-priced at the cheapest nearby store.
HIGHLIGHTS
- Geospatial queries in PostGIS: stores within a 2 km radius, and duplicate stores caught within 50 m.
- Row-level security in Postgres: guests can only read, registered users can contribute, and the UI mirrors the same rule.
- Money stored as whole centavos end to end, and medians use percentile_disc so every price shown is one somebody actually saw.
- 16 versioned schema migrations with test scripts; nothing is ever deleted, flags and corrections are recorded instead.
- Every screen has a loading, empty and error state, designed first as 39 Figma frames.


