Media · Applied AI
ViralClip — live Twitch viral moment clipping
An AI pipeline that watches live streams, scores viral moments, and generates shareable clips for streamers and clip channels.
- Source
- Live Twitch
- Pattern
- Detect → clip
- API
- FastAPI
- Storage
- Postgres + Mongo
Problem
Clip channels and streamers waste hours scrubbing VODs for highlight moments. Manual clipping does not scale with concurrent live streams.
They needed automated detection plus a clip generation pipeline with APIs that could plug into existing creator workflows.
Approach
FastAPI service monitoring Twitch streams, scoring candidate viral moments, and kicking off clip generation jobs.
Dual storage: PostgreSQL for relational session/clip state and MongoDB for higher-throughput document workloads.
API-first design so detection and clip status can be consumed by dashboards or downstream tools.
Outcome
Shipped an AI clipping system that turns live-stream signals into shareable clips — a concrete media/AI delivery pattern beyond chat wrappers.
Stack
- FastAPI
- Python
- PostgreSQL
- MongoDB
- Twitch APIs