Real-Time Live Game Data Tracker — Node.js / Headless Browser Developer

We run a statistics website for live casino game shows. We need a developer to build a real-time data tracker for two specific live game-show tables (Evolution Gaming): capture each round's outcome as it happens and feed it to our website's stats pages (results history, bonus frequency, "drought" trackers, top multipliers).

Most of our platform is already built (backend endpoints, cache, front-end UI, SEO). The core of this job is the data-capture layer — a reliable worker that reads live round results and posts them to our ingestion endpoint.

What you'll build

A 24/7 worker that captures live round outcomes for the two tables in real time
Normalize each outcome into our JSON format and POST it to our secured ingestion endpoint
Robust session handling: keep a persistent live game session alive, auto-reconnect on drops, auto-restart on crash
Monitoring/alerting when the feed stops (so it never fails silently)
Deployment (Docker / VPS / Fly.io) + documentation

One table has a relatively simple data path (a pollable feed). The other is harder: the data lives inside a cross-origin iframe WebSocket and must be captured with a headless browser driving a live game session. Please quote these two separately.

Requirements

Strong Node.js
Proven experience with Puppeteer / Playwright + stealth, headless automation, and persistent authenticated sessions
Experience reverse-engineering WebSocket feeds and handling cross-origin iframes
Comfortable keeping long-running workers stable in production (reconnect, keep-alive, monitoring)
Basic backend (PHP or Node) for the ingest/cache glue

Nice to have

Prior work with live casino / game-show data or similar real-time feeds
Anti-bot / bot-protection experience

Deliverables

Working worker(s) posting live rounds to our endpoint
Monitoring + alerting
Full source code + deployment README
Clear docs on how the sources are configured and how to refresh a session

Acceptance criteria

Rounds appear on our pages within ~10 seconds of the real round
The hard table's worker runs stable for 72+ hours with no manual intervention
Clean "offline" state when a table isn't broadcasting; automatic recovery

To apply, tell us:

Have you built real-time scrapers for live game feeds or WebSockets? Share examples.
How do you keep a headless browser session alive 24/7 and handle disconnects?
How do you capture a WebSocket inside a cross-origin iframe?
How do you monitor a long-running worker and alert on failure?
Your estimate (time + price) — split into the two tables separately.