BUILTPrototype

LocalLoco

Neighbourhood deal discovery, sourced by residents rather than a sales team.

My role: Full-stack developer and co-founder
Timeframe: September 2024 – present
Status: Prototype
Location: Singapore

The problem

Small neighbourhood businesses rarely have time to keep listings current on the big platforms, so their promotions stay invisible outside their own shopfront. LocalLoco inverts that: residents submit, volunteers moderate, and QR redemption gives the business a real signal. The hard parts were keeping location queries fast and keeping the moderation queue small enough for a few people to clear.

I owned1/4
- Split a Next.js front end from a FastAPI service so each deploys independently
- Modelled and built the location query layer in PostgreSQL
- Resident submission flow and the volunteer moderation queue behind it
- QR generation and redemption tracking — redemptions, not impressions
- Mobile-first responsive interface
Constraints2/4
- Content quality depends entirely on volunteers
- Almost all real use is on a phone, often on mobile data
- Location queries degrade fast if modelled naively
- No marketing budget — it has to be useful on first visit
Outcome3/4
- Moved from a throwaway v0 preview to the production domain (www.localloco.sg), with real auth middleware wired up
- Location queries served through PostGIS rather than application-side filtering, which is what keeps them viable as data grows
- Submission and moderation are separate stages, so unreviewed content is never public
- Redemption tracking gives businesses a real signal instead of a vanity metric
Learned4/4
- A community catalogue lives or dies on moderation throughput, not on the submission form
- Choosing PostGIS early avoided a painful rewrite once real data existed
- I over-invested in the business analytics view before there were enough redemptions to make it meaningful