### 1. System Architecture Overview The system employs a multi-tenant microservices-lite architecture: * **Presentation Tier:** Next.js 15 hosted on Edge CDN with Server-Side Rendering (SSR) for initial loads and static asset distribution. * **API Gateway & Routing:** AWS Application Load Balancer (ALB) directing requests to containerized FastAPI services running on AWS ECS Fargate. * **Tenant Isolation:** Row-Level Security (RLS) within PostgreSQL mapped to a unique `tenant_id` on all tables, enforced by JWT middleware. * **Asynchronous Processing:** Redis-backed Celery workers executing long-running export and ingestion tasks out-of-band. ### 2. Database Schema & Data Model * **tenants:** `id` (UUID, PK), `company_name` (VARCHAR), `plan_tier` (VARCHAR), `created_at` (TIMESTAMP) * **users:** `id` (UUID, PK), `tenant_id` (UUID, FK -> tenants.id, INDEX), `email` (VARCHAR, UNIQUE), `password_hash` (VARCHAR), `role` (VARCHAR) * **leads:** `id` (UUID, PK), `tenant_id` (UUID, FK -> tenants.id, INDEX), `assigned_user_id` (UUID, FK -> users.id), `status` (VARCHAR), `deal_value` (NUMERIC(12,2)), `created_at` (TIMESTAMP, INDEX) * **activity_logs:** `id` (UUID, PK), `tenant_id` (UUID, FK), `lead_id` (UUID, FK), `event_type` (VARCHAR), `payload` (JSONB) * **Indexing Strategy:** Compound index on `(tenant_id, status)` and `(tenant_id, created_at DESC)` for sub-millisecond filtering. ### 3. Core API Contract Specifications | Method | Route | Request Payload Structure | Response Codes | | :--- | :--- | :--- | :--- | | **POST** | `/api/v1/auth/token` | `{"email": "user@org.com", "password": "..."}` | `200 OK`, `401 Unauthorized` | | **GET** | `/api/v1/leads` | Query params: `?status=open&page=1&limit=25` | `200 OK`, `403 Forbidden` | | **POST** | `/api/v1/leads` | `{"name": "Acme Corp", "value": 50000, "assigned_to": "UUID"}` | `201 Created`, `422 Unprocessable` | | **PATCH** | `/api/v1/leads/{id}/status`| `{"status": "negotiation", "notes": "Demo booked"}` | `200 OK`, `404 Not Found` | ### 4. Cloud Infrastructure & CI/CD Pipeline * **Compute:** AWS ECS Fargate with auto-scaling triggers (CPU > 70% or Memory > 75%, min 3, max 20 tasks across 3 AZs). * **Database:** Amazon RDS Aurora PostgreSQL with 1 Writer node and 2 Read Replicas across multi-AZ configurations. * **Caching Layer:** Amazon ElastiCache (Redis cluster mode enabled) for session store and hot tenant data caching. * **CI/CD Pipeline:** GitHub Actions running linting, PyTest unit suites, Docker image generation, pushing to AWS ECR, and initiating zero-downtime rolling ECS deployments. ### 5. Security & Performance Optimization * **Authentication:** Stateless asymmetric RS256 JWT tokens containing `tenant_id`, `user_id`, and `role` claims with 15-minute expirations and Redis refresh token rotation. * **Rate Limiting:** Token-bucket rate limiter via Redis returning `429 Too Many Requests` (100 req/min per tenant). * **Caching Topology:** Read-through caching pattern on dashboard metrics with a 60-second TTL to offload Aurora read replicas.