Skip to content
Mahavir

← All work

Codebasics.io

2020 – Present · AtliQ Technologies · Full-stack owner — performance, payments, video pipeline, observability

Users
1M+
Uptime
99%+
Cost cut
60%
Query speedup
40x

Highlights

  • Scaled to 1M+ users at 99%+ uptime with Multi-AZ RDS and CloudFront
  • Cut cloud spend 60% via build-vs-buy overhaul (self-hosted Supabase)
  • Critical stored procedure 4s → 100ms; overall DB performance +30%
  • Multi-currency payments with idempotent transactions and webhook reconciliation
  • Cross-provider video auto-sync with retries, failed-sync recovery, and DRM delivery

Overview

Codebasics.io is a high-traffic B2B/B2C learning platform I’ve owned full-stack for years. It serves 1M+ users at 99%+ uptime, handling multi-currency payments, role-based portals, a complex video pipeline, and the operational reliability that scale demands.

This is the project where I learned that “working” and “working in production at scale” are different problems — and that the only way to find real bottlenecks is to trace production behaviour.

The challenge

A data-science & analytics LMS that needed to scale to millions of users across regions, handle real money in multiple currencies, deliver video smoothly, and stay up under load. The cost of getting any of this wrong — failed payments, broken video, downtime during a launch — was high.

What I built & scaled

Performance & infrastructure

  • AWS Elastic Beanstalk for deployment with Multi-AZ RDS for database resilience.
  • Redis caching strategy that took hot paths off the database.
  • Database indexing and query optimisation — most notably taking a critical stored procedure from 4 seconds down to 100ms, lifting overall database performance by 30%.
  • Laravel Octane to keep PHP workers alive and cut per-request overhead.
  • Nginx tuning and CloudFront for global content delivery.
  • Self-hosted Supabase as part of a build-vs-buy overhaul that cut cloud spend by 60%.

Multi-currency payments

Real money, multiple currencies, multiple gateways. I built idempotent transaction handling so retries never double-charge, plus webhook-driven reconciliation across Stripe and Authorize.Net — so the system’s view of “what was paid” always matches the payment provider’s view.

Video pipeline

A video pipeline with cross-provider auto-sync, retries, failed-sync recovery, and monitoring — plus DRM delivery (Gumlet, with Cloudflare as fallback) so content is protected end-to-end. Course-progress tracking ties it all back to the learner.

Production observability

This is where the real work happened. New Relic and CloudWatch to trace production behaviour and find actual bottlenecks (not the ones you’d guess). Uptime Kuma for monitoring. Retries and reconciliation so transient failures self-heal. The result: 99%+ uptime at 100K+ concurrent users.

Why it matters

Codebasics.io is a textbook example of owning a product at scale: the architecture, the cost discipline, the operational reliability, and the production-tracing mindset that finds the one stored procedure killing your performance. It’s been in production, serving real users, for years.

My role

Full-stack ownership — performance tuning, payments, video, infrastructure decisions, and the production observability that kept it healthy.