Cash Management Command Center
Cloud Infrastructure
Problem & Solution
Enterprise retail owners with multiple locations face a 'blind spot' in their most critical asset: physical cash. Without centralized visibility, monitoring cash flow across distributed branches becomes a manual, error-prone nightmare. Fragments of data across different sites make it impossible to detect fraud, predict cash replenishment needs, or monitor the mechanical health of expensive automated teller machines in real-time.
I architected and built a centralized FinTech Cloud Platform that serves as the mission-control for multi-shop cash management. The platform acts as a high-availability ingestion engine for Cash Management Local Systems, aggregating machine-level data—down to individual bill denominations—into a unified dashboard. By implementing secure API key management and real-time WebSocket synchronization, I transformed fragmented local operations into a data-driven, transparent ecosystem that gives owners total control over their cash inventory from anywhere in the world.
Key Features & Business Impact
- Real-Time Denomination Tracking: Deep visibility into specific bill and coin counts across all connected machines.
- Hardware Health Telemetry: Remote monitoring of Glory Cash Infinity status, providing proactive alerts for maintenance.
- Unified API Key Management: Secure, dashboard-controlled key rotation and revocation for all remote branch systems.
- Audit-Ready Transaction History: Secure storage of 3 months of granular transaction data with complete cryptographic integrity.
- Multi-Shop Dashboard: Aggregated financial metrics and cash-flow trends across multiple geographic locations.
- Idempotent Data Sync: Robust synchronization logic ensuring zero duplicate records and zero data loss during connectivity flickers.
Technical Deep Dive
Cloud-Native Multi-Tenant Architecture
Cweetlabs is built on a high-concurrency Node.js backend optimized for multi-tenant isolation. The platform utilizes NGINX as a high-performance reverse proxy for SSL termination and request routing, with PM2 managing a cluster of Node.js instances to ensure zero-downtime deployments. This architecture is designed to scale horizontally, supporting hundreds of concurrent shop locations while maintaining low-latency API response times for critical hardware synchronization events.
Security: JWT, API Key Management & RBAC
Security is paramount when handling financial data. I implemented a dual-layer authentication system: Managed API Keys with rotation capabilities for hardware-to-cloud communication, and Short-Lived JWTs for web-based administrative access. The system enforces strict RBAC (Role-Based Access Control), ensuring that branch-level employees only interact with their assigned machines, while global administrators maintain oversight across the entire organization.
Ingestion: Real-Time Sync & Idempotency
To ensure 100% data accuracy in financial records, the data pipeline utilizes Idempotency Keys to prevent duplicate transaction entries during network retries. Local systems transmit transaction batches (deposits, withdrawals, and denominations) which are validated and processed asynchronously. I implemented a WebSocket-driven Event Bus to push live transaction alerts to the dashboard, providing shop owners with sub-second 'Live View' of their cash inventory across all global sites.
Scalability: MongoDB Aggregation & TTL Strategy
The database layer is optimized for high-write workloads and complex analytical queries. I utilized MongoDB Compound Indexes on (shop_id, created_at) to accelerate historical reporting. To maintain peak performance, I implemented an Automatic Data Lifecycle Strategy using TTL (Time-To-Live) Indexes for raw logs, while preserving aggregated financial summaries in materialized collections. Complex denomination-level breakdowns are handled via optimized Aggregation Pipelines, allowing for rapid audit-trail retrieval without impacting the primary ingestion flow.
Technology Stack
NextJS
Typescript
TailwindCSS
MongoDB
NGINX
PM2