AbdulmalikAlayande/amala-atlas-backend-api

Django REST API powering Amala Atlas — handles multi-channel candidate ingestion (WhatsApp, web, Google Maps, Twitter), source-channel-aware scoring, fuzzy deduplication, dynamic verification thresholds, and promotion to canonical map database. The system of record for every Amala spot. # Amala Atlas — Backend API The **system of record** for the Amala Atlas ecosystem. Manages the full lifecycle of an Amala spot — from raw submission through scoring, deduplication, verification, and promotion to the canonical map database. ## Architecture The backend is a Django project with six apps, each handling a distinct domain: ``` amala_atlas/ # Django project config (settings, urls, wsgi) ├── places/ # Core: Spot, Candidate, Submission, PhotoURL │ ├── models.py # Data models │ ├── services.py # Scoring, geocoding integration, deduplication │ ├── serializers.py # DRF serializers │ ├── views.py # Spot API + candidate submission endpoint │ ├── filters.py # Bbox, city, tag, query filters for spots │ ├── nlp_utils.py # Nigerian food keywords + phone extraction │ ├── geocoding.py # Nominatim + 14-city centroid fallback │ └── dedupe.py # Name normalization, phone matching, fuzzy dedup ├── verification/ # Human review pipeline │ ├── models.py # Verification model (approve/reject/merge) │ └── views.py # Queue, action endpoint, dynamic thresholds ├── ingestion/ # Automated discovery intake │ ├── views.py # POST /ingest/ and /ingest/batch/ (API key auth) │ ├── extractors.py # HTML content extraction utilities │ └── agents/ # Discovery agents │ ├── google_maps.py # Google Places API scanner │ └── twitter.py # Twitter/X venue mention extractor ├── whatsapp/ # WhatsApp bot via Twilio │ ├── models.py # WhatsAppSession (conversation state) │ ├── conversation.py # State machine: awaiting_spot → ... → complete │ ├── parser.py # Nigerian spot name/location text parsing │ └── views.py # Twilio webhook endpoint ├── users/ # Custom User with WhatsApp phone + contribution tracking └── com …