Projects Archive

Umedmi

A patient management platform built from scratch — schema design, API, AWS infrastructure, real-time chat and payment integration. Three years on a production healthcare system.

Visit site
Umedmi

Category

Full-Stack

Year

2022

Duration

2+ years

Team

5 engineers

Overview

  • Built a comprehensive patient management system from the ground up — defined the data model, designed the API layer, and set up the cloud infrastructure for a healthcare platform handling sensitive medical data.
  • Designed and implemented an ICD-10 medical code parsing engine that transformed complex diagnostic codes into searchable, interactive cards. Reduced clinical data lookup time for healthcare providers meaningfully.
  • Designed and managed the full AWS production infrastructure (EC2, S3, IAM, RDS) — security groups, automated backups, CloudWatch monitoring, HIPAA-aligned deployment practices.
  • Co-architected a real-time secure chat system using WebSockets with end-to-end message handling and data privacy safeguards.
  • Built a multi-provider payment integration with automated reconciliation — secure transactions across providers with audit trails for every transaction.

The problem

  • Healthcare platforms in this space either lacked proper data standards (ICD-10 compliance) or were too expensive for small clinics to adopt.
  • Medical code lookup was manual and slow — clinicians were searching through PDFs and reference sheets to find the right diagnostic code.
  • No real-time communication between clinic staff and patients within the platform — everything was phone-based.
  • AWS infrastructure was initially unmanaged with no backups, monitoring, or access controls — a liability for a platform holding medical records.

The solution

  • Designed a normalized relational schema that could represent patient records, appointments, medical codes, and billing in a way that was both queryable and auditable.
  • Built an ICD-10 parser that ingested the full code set, indexed it for search, and rendered each code as a card with description, category, and related codes — full-text search across 70k+ codes.
  • Set up a proper AWS production environment: VPC with private subnets, EC2 in an ASG, RDS with automated snapshots, CloudWatch alarms, and IAM roles with least-privilege access.
  • Implemented WebSocket-based chat with message persistence, read receipts, and encryption at rest — built so clinic staff could message patients directly from the platform.
  • Integrated multiple payment gateways with a reconciliation layer that ran nightly and flagged any mismatches for manual review.

Outcomes

  • ICD-10 engine indexed 70k+ medical codes with full-text search — lookup time reduced significantly.
  • AWS infrastructure brought to HIPAA-aligned standards with proper IAM, backups, and monitoring.
  • Real-time chat shipped with WebSockets — reduced phone-based coordination overhead for clinic staff.
  • Multi-provider payment reconciliation caught and flagged transaction mismatches automatically.
  • Platform delivered from zero — schema, API, infrastructure, and frontend built by a team of 5.

My role

Full-Stack Developer

TypeScript, JavaScript, Node.js, NestJS, AWS, MongoDB, Git, Redis