Back to projects

TelcoLearn — LMS & Online Labs

A learning platform where students buy courses and spin up live, cloud-provisioned lab machines on demand.

Live site

My role: Full-Stack Engineer — built the Next.js client and the Express/MongoDB API end-to-end.

~51.5k
Frontend LOC
474
Frontend files
25
Backend models
~111
API endpoints

Overview

TelcoLearn is a full learning platform: a course catalog, a Notion-style authoring experience, Razorpay checkout with GST invoicing, and an RBAC admin panel — plus the standout feature, on-demand online labs. When a student starts a lab, the backend provisions a real AWS EC2 instance with a time-to-live, tracks its lifecycle, and an hourly cron reclaims and archives expired instances automatically.

The problem

Hands-on telecom/networking training needs real machines, not videos. Giving every student a persistent server is expensive and insecure. The platform had to provision isolated lab environments on demand, enforce time limits, and tear everything down reliably — all wrapped in a normal e-commerce + LMS experience with payments, progress tracking, and certificates.

What I built & the hard parts

On-demand cloud labs with TTL lifecycle

A lab purchase provisions an AWS EC2 instance (id, public/private IP, type, state) tracked in UserLabInstance with a time-to-live. A node-cron job runs hourly to terminate expired instances and move them to an archive collection — so compute is never left running and paid for.

Granular RBAC admin panel

Role + Permission models give module-by-action control (create/read/update/delete per module). Super admins bypass; everyone else is checked by a hasPermission middleware. The admin UI manages courses, labs, users, blogs, coupons, and roles.

Payments done properly

Razorpay with HMAC-SHA256 signature verification, multi-item carts, 18% GST tax invoices, financial-year payment numbering, and a 30-day refund window with reason tracking.

Notion-style authoring + resumable video

The frontend uses Plate.js (24+ plugins) for rich course/blog content. Videos upload to Vimeo via resumable 5MB chunked uploads with exponential-backoff retry, keeping large uploads reliable on flaky connections.

Architecture

Online lab lifecycle

From purchase to automatic reclamation — the core differentiator of the platform.

System architecture

Next.js client, Express API, MongoDB, and an external lab-provisioning service in front of AWS.

Features

Learning

  • ›Course → section → chapter structure
  • ›Vimeo-hosted video lessons
  • ›Progress tracking + certificate generation
  • ›Course & lab ratings / reviews

Online labs

  • ›On-demand AWS EC2 provisioning
  • ›TTL-based access windows
  • ›Hourly cron lifecycle + archival
  • ›Per-user instance dashboards

Commerce & admin

  • ›Razorpay + GST invoices + refunds
  • ›Multi-item cart & coupons
  • ›RBAC admin panel (module × action)
  • ›Multi-provider auth (JWT, Google, GitHub)

Tech stack

Frontend
Next.js 15
React 19
TypeScript
Tailwind + shadcn/ui
Redux Toolkit
Plate.js
Backend
Express
MongoDB
Mongoose
Passport.js
node-cron
Integrations
AWS EC2
Razorpay
Vimeo
Cloudinary
Docker

Outcome

  • ✓Shipped a production LMS with a genuinely differentiated online-labs feature.
  • ✓Reliable, self-cleaning cloud-lab provisioning that controls compute cost.
  • ✓Complete commerce flow: cart, payments, GST invoicing, refunds, coupons.
  • ✓Enterprise-grade RBAC with fine-grained, per-module permissions.