Amala Atlas

A community-driven location intelligence platform for discovering, mapping, and verifying Amala food spots across Nigeria. We specialize in finding obscure, undocumented, and informal locations that fly under the radar of mainstream directories like Google Maps.

The Problem

Nigeria's best Amala spots are invisible online. They don't have websites, Google listings, or social media pages. They're known by word of mouth: "Iya Basira under the bridge in Mokola" or "that buka behind the market in Surulere." This knowledge lives in people's heads and WhatsApp groups — not on the internet.

The Solution

Instead of scraping the web for data that doesn't exist, we go where the knowledge is: the people who eat Amala every day. Amala Atlas turns community knowledge into verified map data through multiple discovery channels, all feeding into a single verification pipeline.

How It Works

User eats Amala  ->  Their spot is missing from the map  ->  They submit via WhatsApp/web
     -> Spot gets verified  ->  Appears on the map  ->  User shares "I added my spot"
     -> Friends see it  ->  Their spots are missing  ->  They submit  ->  Cycle repeats

System Architecture

This project is organized as a monorepo hub with three components:

amala-atlas/
├── amala-atlas-backend-api/     # Django REST API — the system of record
├── amala-atlas-auto-discovery/  # Scrapy-based web crawler (archived)
└── amala-atlas-explorer/        # React frontend (separate repo)

Component	Role	Tech	Repo
Backend API	System of record — ingestion, verification, spot lifecycle	Django 5.x, DRF, PostgreSQL	amala-atlas-backend-api
Auto-Discovery	Web crawler for bootstrapping data (archived — replaced by lightweight agents in the backend)	Scrapy, spaCy, SQLAlchemy	amala-atlas-auto-discovery
Frontend	Map UI for browsing, searching, and submitting spots	React, Vercel	amala-atlas-explorer

Discovery Channels

Data enters the platform through multiple channels, each with a different trust level:

Channel	How It Works	Trust Level	Verification Needed
WhatsApp Bot	Users text a spot name + location pin to a WhatsApp number	High (GPS proof)	1 approval
Web Form	Users submit via the frontend at amalatlas.vercel.app	Medium	1-2 approvals
Google Maps Agent	Automated scan of Google Places API for amala-related searches	High (Google-reviewed)	1 approval
Twitter Agent	Monitors tweets mentioning amala spots with city context	Low	2 approvals
Seed Script	One-time bootstrap from Nigerian food blogs	Low	2 approvals

Candidate Lifecycle

Every discovered spot follows the same pipeline regardless of source:

Submission  ->  Candidate (scored + deduplicated)  ->  Verification Queue  ->  Spot (on the map)

1. Submission: Raw data enters the system (name, address, coordinates, photos)
2. Candidate: Scored based on evidence quality and source channel trust (0.0–1.0)
3. Verification: Human reviewers approve/reject with dynamic thresholds per channel
4. Spot: Promoted to the canonical map database

Getting Started

Each component has its own setup instructions:

• Backend API: See the Backend README
• Auto-Discovery (archived): See the Auto-Discovery README
• Frontend: See the Explorer repo

Quick Start (Backend Only)

cd amala-atlas-backend-api
pip install -r requirements.txt   # or: pipenv install
cp .env.example .env              # configure DATABASE_URL, SECRET_KEY
python manage.py migrate
python manage.py runserver

Deployment

• Backend: Render (Django + PostgreSQL)
• Frontend: Vercel (React)

Contributing

Contributions are welcome. If you know Amala spots that aren't on the map, the fastest way to contribute is to submit them through the platform.

For code contributions, pick a component repo, read its README, and open a PR.

---

Amala Atlas — Mapping the flavor of Nigeria, one spot at a time.