Real estate listing and mortgage pipeline platform. Search properties, schedule showings, negotiate offers, and track mortgage applications — end to end.
Live demo → parcela-zeta.vercel.app
Buyers
- Full-text and geo-radius property search (Elasticsearch)
- Schedule and manage showings
- Submit offers with a 3-step modal — price, contingencies, review
- Counter-offer chain with earnest money and closing date negotiation
- Mortgage application pipeline with document upload and OCR parsing
Agents
- Create and publish listings with Cloudinary image uploads
- CRM dashboard: Kanban board across deal stages, today's showings, pending offers
- Real-time offer negotiation via Socket.io
Admin
- Platform stats, BullMQ queue health monitoring, user management
graph TB
subgraph "Vercel"
Web["Next.js 15\nApp Router"]
end
subgraph "Fly.io"
API["Fastify API\n(REST + WebSocket)"]
Workers["BullMQ Workers\n× 5"]
Redis[("Redis")]
end
subgraph "Neon"
DB[("PostgreSQL\n15 tables")]
end
subgraph "AWS"
S3[("S3\nImages & Docs")]
Textract["Textract OCR"]
end
ES[("Elasticsearch\nListings index")]
Web -->|"REST / Socket.io"| API
API --> DB
API --> Redis
API --> ES
API -->|"Presigned URLs"| S3
Workers --> DB
Workers --> ES
Workers --> Textract
Workers --> Redis
sequenceDiagram
participant Buyer
participant API
participant Agent
Buyer->>API: POST /listings/:id/offers
API-->>Agent: offer.submitted (Socket + email)
Agent->>API: POST /offers/:id/counter
API-->>Buyer: offer.countered (Socket + email)
Buyer->>API: POST /offers/:id/accept
API-->>Agent: offer.accepted (Socket + email)
API->>API: Create transaction record
stateDiagram-v2
[*] --> pre_qualification
pre_qualification --> application_submitted
application_submitted --> processing
processing --> conditionally_approved
processing --> denied
conditionally_approved --> clear_to_close
clear_to_close --> closed
closed --> [*]
denied --> [*]
parcela/
├── apps/
│ ├── api/ # Fastify backend — raw SQL, BullMQ, Socket.io
│ └── web/ # Next.js 15 — React Query, Zustand, Mapbox
├── packages/
│ └── shared-types/ # Shared TypeScript interfaces
└── docs/ # Architecture and API reference
Prerequisites: Node 20+, pnpm 9+, Docker (for Elasticsearch)
# Install
git clone https://github.com/StephaneWamba/parcela.git
cd parcela && pnpm install
# Start Elasticsearch locally
docker compose up -d
# Copy env files and fill in values
cp apps/api/.env.example apps/api/.env
cp apps/web/.env.local.example apps/web/.env.local
# Run database migrations
pnpm --filter @parcela/api db:migrate
# Start development servers
pnpm dev
# → API on http://localhost:3000
# → Web on http://localhost:3001| Variable | Purpose |
|---|---|
DATABASE_URL |
Neon PostgreSQL connection string |
REDIS_URL |
Redis (BullMQ + idempotency) |
ELASTICSEARCH_URL |
Elasticsearch endpoint |
JWT_ACCESS_SECRET |
32+ char access token secret |
JWT_REFRESH_SECRET |
32+ char refresh token secret |
AWS_ACCESS_KEY_ID / AWS_SECRET_ACCESS_KEY |
S3 + Textract |
S3_BUCKET |
Bucket for documents and images |
CLOUDINARY_CLOUD_NAME / CLOUDINARY_API_KEY / CLOUDINARY_API_SECRET |
Listing images |
RESEND_API_KEY |
Transactional email |
GOOGLE_CLIENT_ID / GOOGLE_CLIENT_SECRET |
OAuth2 |
See apps/api/.env.example for the full list.
| Variable | Purpose |
|---|---|
NEXT_PUBLIC_API_URL |
Backend URL |
NEXT_PUBLIC_MAPBOX_TOKEN |
Mapbox GL map |
API → Fly.io
flyctl deploy --config apps/api/fly.toml --remote-onlyWeb → Vercel
npx vercel --prod --yesDatabase migrations are run manually against Neon branches. Dev branch is pre-seeded with 20 Bay Area listings, agents, and a 5-round offer counter chain.
- Architecture — services, data flow, state machines
- API Reference — all endpoints with request/response shapes
- Database Schema — table definitions, indexes, relationships
MIT