Sathiyamoorthi Madhesh
2026Document Retrieval

RCM DRS 

Bulk claim-document retrieval from the DHPO transaction bus, with live progress that survives a closed browser tab.

RCM DRS — project preview
Role
Full-Stack Developer
Timeline
Feb — Jun 2026
Stack
  • Angular 18
  • Angular Material
  • Node.js
  • Socket.IO
  • Worker Threads
  • MongoDB

Pulling documents out of DHPO one at a time is fine until someone asks for a month of transactions. DRS turns that into a batch: search, select, download, and get a zip — while a Socket.IO room streams incremental progress to whoever is watching, so a long-running job reports itself instead of appearing to hang. Rooms are addressable and resumable, so closing the tab does not kill the download.

The bottleneck was never the network, it was parsing. Every retrieved transaction is XML, and doing that on the event loop stalls every other request on the box. Parsing runs in a worker-thread pool sized to available cores, with workers recreated automatically on crash or non-zero exit so a single malformed document cannot drain the pool.

The rest is the unglamorous half that makes it deployable inside a hospital network: JWT with refresh-token rotation, optional TOTP two-factor, rate limiting, session logging, role-scoped user and client management, and a configurable outbound proxy for the many environments where nothing reaches the internet directly.

Realtime
Batch progress
Worker pool
XML parsing
JWT + 2FA
Auth