Scaling PomPak to 1M+ Users
Engineering a highly available national financial literacy platform backed by the State Bank of Pakistan. How I led the backend to support massive concurrent spikes across 45+ districts.
The Challenge
The platform faced extreme traffic patterns typical of national rollout programs: massive, unpredictable spikes during school hours when entire districts logged on simultaneously to take financial literacy quizzes. The existing monolithic approach was struggling under database transaction locks, causing dreaded timeout drops during high-stakes exams.
The Architecture
Visual Deep Dive
PomPak Learning Platform
To solve the concurrency bottlenecks, I engineered a massive overhaul:
- SQL Server Stored Procedures: Offloaded complex grading analytics and multi-table joins from Laravel's Eloquent ORM directly into raw SQL stored procedures, reducing execution time by 85%.
- Stateless Containerization: Dockerized the Laravel application to allow AWS EC2 instances to auto-scale horizontally when CPU load breached 70%.
- Optimistic Caching: Integrated caching for high-frequency low-mutation routes.
The Result
PomPak scaled beautifully during the national phase-out. Not only did we achieve over 1 million users reliably, but infrastructure costs scaled efficiently, and server timeout metrics plummeted to absolute zero during examination hours. This project cemented my conviction that framework abstractions are a tool, but database mastery is fundamental.