Looking for Devs: Real-Time Automation System for Live Streaming Platforms (AI + OBS + Node.js)
Budget: $30 – $250 USD
I’m looking for a compact proof-of-concept that proves we can automate a live-cam workflow end-to-end. Long-term the system will span seven adult streaming sites, but for this first milestone I only need a minimal, working slice so we can validate the tech stack on a single platform of your choice.
Scope for this iteration
• Launch a headless browser with Puppeteer / Playwright, join one live room, and capture both chat messages and tip events in real time.
• Pipe each event to a Node.js service on my Linode Ubuntu box, normalised as JSON and broadcast over WebSocket (expose a simple REST endpoint for health-check).
• The Node service triggers an overlay change in OBS or vMix via the OBS WebSocket API—show me any visual cue that a tip was received.
• Persist events to a structured log (JSON lines file or lightweight database) so we have a replayable history.
• Docker-compose everything: headless browser, Node backend, and an nginx reverse proxy with basic TLS so I can spin it up with one command.
• Include a stub endpoint for future OpenAI/LLM replies (return a hard-coded message for now).
• Provide a bare-bones React dashboard that streams the WebSocket feed live and lists the last 100 events.
Tech requirements
Node.js, WebSocket, Puppeteer / Playwright, OBS WebSocket API, Docker, Linux (Ubuntu). Keep the code clean, documented, and secure—no hard-coded credentials.
Delivery
Git repo + docker-compose.yml + short README showing how I start the stack and see:
1) live console logs in the backend,
2) an overlay change in OBS, and
3) events streaming in the React page.
If this demo is solid, I’ll open follow-up work to add the remaining six platforms, full conversational AI, role-based admin panel, and advanced analytics.
Scope for this iteration
• Launch a headless browser with Puppeteer / Playwright, join one live room, and capture both chat messages and tip events in real time.
• Pipe each event to a Node.js service on my Linode Ubuntu box, normalised as JSON and broadcast over WebSocket (expose a simple REST endpoint for health-check).
• The Node service triggers an overlay change in OBS or vMix via the OBS WebSocket API—show me any visual cue that a tip was received.
• Persist events to a structured log (JSON lines file or lightweight database) so we have a replayable history.
• Docker-compose everything: headless browser, Node backend, and an nginx reverse proxy with basic TLS so I can spin it up with one command.
• Include a stub endpoint for future OpenAI/LLM replies (return a hard-coded message for now).
• Provide a bare-bones React dashboard that streams the WebSocket feed live and lists the last 100 events.
Tech requirements
Node.js, WebSocket, Puppeteer / Playwright, OBS WebSocket API, Docker, Linux (Ubuntu). Keep the code clean, documented, and secure—no hard-coded credentials.
Delivery
Git repo + docker-compose.yml + short README showing how I start the stack and see:
1) live console logs in the backend,
2) an overlay change in OBS, and
3) events streaming in the React page.
If this demo is solid, I’ll open follow-up work to add the remaining six platforms, full conversational AI, role-based admin panel, and advanced analytics.