# Video pipeline on AWS

**Result:** 60% less manual editing  
**Category:** Cloud  
**Stack:** AWS, AWS Lambda, SQS, S3, Node.js, React

A serverless pipeline for a crowdsourced video platform: contributors record on their phones, and uploads are transcoded and assembled into branded videos without anyone moving files by hand. Built for ShoutOut.

## The problem

Every crowdsourced submission needed manual uploading, transcoding, review and assembly. It worked for a few hundred users and broke down beyond that, and the team spent more time moving files than improving the product.

## Results

- **60%** less manual work
- **1,000+** active users
- **0** servers to manage

## What I did

1. **Event-driven.** S3 uploads trigger Lambda functions, orchestrated through SQS.
2. **Async processing.** Heavy transcoding runs in background workers, so the app stays responsive under load.
3. **Zero ops.** No servers to manage: it scales with traffic and idles near zero cost between bursts.
4. **Durable state.** Job state, automatic retries and dead-letter handling, so failures are never silent.
5. **Live dashboard.** A React dashboard shows each submission’s status at every stage.

## What it does

- Video templates with intro, contributor and outro segments
- Contributor invites by QR code and link
- Record from a phone, no app needed
- Clips, photos, playlists and compilations
- Live status for every submission

> “His communication was clear, regular and fast.” John Greenwood, Tech lead, ShoutOut

Screenshots are blurred and anonymised. Page: https://venuspaneliya.com/work/video-pipeline-aws/