Oladele Israel
← All work

Moola Business Banking

High-throughput transaction processing engine powering Nigerian banking systems

Type
Company Product
Year
2025
Context
Virtual Switch INTL
Role
Backend Software Engineer
Status
Live
Visit the live site ↗
Moola Business Banking screenshot

Overview

Moola Business Banking is a high-availability banking platform engineered for business and enterprise account operations. As part of Virtual Switch INTL, I developed the backend pipeline handling high-volume transaction processing, real-time account ledgers, automated loan disbursements,saving circle operations security validation, and automated compliance logging.

Why I built it

Financial systems in Nigeria demand absolute uptime and unyielding precision under heavy concurrent workloads. Building backend infrastructure for personal and Business banking at Virtual Switch INTL allowed me to solve complex distributed transaction problems where performance and correctness cannot be compromised.

Stack

  • Node.js & NestjsAsync I/O non-blocking runtime ideal for processing thousands of concurrent banking webhooks and transaction events.
  • PostgreSQLACID-compliant relational storage ensuring ledger balance integrity across complex debit and credit operations.
  • RedisDistributed lock manager and high-speed cache for idempotency checking and preventing double-spend risks.
  • AES-256 & Web CryptoPayload encryption for payload payloads and compliance with financial security directives.

Problems & fixes

Problem

Network retries from client apps occasionally triggered duplicate transaction processing during high-latency periods.

What I did

Implemented a distributed Redis idempotency key middleware with atomicity guarantees, ensuring every transaction is executed exactly once.

Problem

Database connection pool starvation during peak business hours transaction surges.

What I did

Optimized query patterns with read/write splitting and configured dynamic connection pooling with query timeout boundaries.

What I learned

  1. Idempotency is non-negotiable in financial systems — every endpoint must handle retries safely without duplicate state mutations.
  2. Asynchronous background job queueing is essential for offloading compliance logging, webhooks, and heavy reconciliation workflows off the main request-response path.
  3. Database model development & query optimization — strategic indexing, normalized ledger schemas, and query tuning prevent lock contention under high concurrency.
  4. Critical thinking & defensive programming — anticipating edge cases, network partitions, and third-party gateway timeouts before code hits production.
  5. Immutable audit trails — recording every balance change with strict ledger history ensures financial accuracy and auditability.
  6. Monitoring transaction latency at p99 reveals bottleneck micro-patterns invisible at p50.
  7. Clean separation between transfer routing logic and ledger balance mutations prevents race conditions.

Next project

Airline Reservation Platform →