Multi-tenant ML-powered default rate prediction API system built with FastAPI.

• Python 3.11+
• Docker & Docker Compose
• Git

1. Clone and setup

   git clone https://github.com/accunodeai/server.git
   cd backend
   make setup

2. Start services

   make start

3. Access the application

   • API: http://localhost:8000
   • Docs: http://localhost:8000/docs
   • Health: http://localhost:8000/health

make help      # Show all available commands
make start     # Start development services
make stop      # Stop all services
make restart   # Restart services
make logs      # View service logs
make test      # Run tests

Multi-tenant SaaS platform with hierarchical organization structure:

• Tenants → Organizations → Users
• Role-based access control: user, org_member, org_admin, tenant_admin, super_admin
• ML Models: Annual (5 ratios) + Quarterly (4 ratios) predictions

• API: FastAPI 2.0 with async PostgreSQL
• Database: PostgreSQL + SQLAlchemy ORM
• Cache: Redis (ElastiCache in production)
• ML: scikit-learn, LightGBM, pandas
• Infrastructure: AWS ECS Fargate, RDS, ALB
• CI/CD: GitHub Actions → ECR → ECS

• Input: Current assets, Total assets, Current liabilities, Total liabilities, Earnings
• Output: Default probability prediction
• Algorithm: Logistic Regression with preprocessing pipeline

• Input: Current assets, Total assets, Working capital, Total debt
• Output: Default probability prediction
• Algorithm: Ensemble (LightGBM + Logistic Regression)

super_admin     → Full system access
tenant_admin    → Tenant-wide management
org_admin       → Organization management
org_member      → Organization data access
user           → Personal data only

• Personal: Own data only
• Organization: Organization-scoped data
• System: Cross-tenant access (super_admin)

# Development
cp .env.example .env.development

# Required variables
DATABASE_URL=postgresql://admin:dev_password_123@localhost:5432/accunode_development
REDIS_URL=redis://:dev_redis_password@localhost:6379/0
SECRET_KEY=your-secret-key

## 🚢 Deployment

### Production (AWS ECS)
- **Trigger**: Push to `prod` branch
- **Pipeline**: GitHub Actions → ECR → ECS Fargate
- **Services**: API + Worker containers
- **Infrastructure**: RDS PostgreSQL + ElastiCache Redis

### Environment Configuration
```bash
# Production secrets via AWS Parameter Store
/accunode/prod/database-url
/accunode/prod/redis-url
/accunode/prod/secret-key

• Complete Docs - Full documentation index
• API Reference - All API endpoints
• System Architecture - Technical design
• Database Design - Schema and relationships