Medtronic Capstone Project

Cloud Infrastructure for Medical Device Monitoring

A complete platform for ingesting device telemetry, analyzing patient data, and providing clinicians with actionable insights—from TimescaleDB to React dashboards.

8 Subsystems
6 Repositories
4 Team Members
MiniMed Pump
Dexcom CGM
FastAPI Auth · Ingest · Read
TimescaleDB
React UI
01

System Architecture

Eight subsystems working together—from device telemetry ingestion through clinical dashboards, with planned AI/ML analytics.

01 Production

Identity & Access

Auth0 OAuth2/OIDC with JWT RS256, role-based access control for patients, clinicians, engineers, and admins.

  • Auth0
  • JWT
  • RBAC
02 Production

Data Ingestion

Bulk import endpoints for CGM readings, pump events, battery voltage, diagnostics, and pressure traces with high-throughput performance.

  • COPY BINARY
  • Pydantic
  • Rate Limiting
03 Production

Data Store

TimescaleDB with hypertables (7-day chunks), row-level security, and continuous aggregates for time-series data.

  • TimescaleDB
  • PostgreSQL 17
  • Hypertables
04 Production

Clinical API

Comprehensive REST API covering patients, glucose data, timeline events, devices, clinical notes, organizations, and administration.

  • FastAPI
  • OpenAPI
  • Async
05 Production

Dashboards

Role-specific React UIs for clinicians, patients, and admins with interactive ECharts visualizations.

  • React 19
  • ECharts
  • Tailwind
06 Partial

Observability

Audit logging middleware, structured JSON logs, and per-request PHI access tracking.

  • Audit Logs
  • JSON Logging
07 In Development

Post-Market Analytics

Cohort analysis, safety signal detection, and regulatory report generation for post-market surveillance.

  • Cohort Builder
  • Safety Signals
08 Planned

AI/ML Analytics

Glucose forecasting, occlusion prediction, anomaly detection, and pressure trace embeddings with pgvector.

  • PyTorch
  • pgvector
  • ONNX
02

Repositories

Six repositories covering the full stack—API, database, UIs, data tools, and analytics.

03

Engineering Rigor

Production-grade practices from day one—security, testing, and performance.

Security First

  • OAuth2/OIDC with MFA support
  • JWT RS256 with JWKS rotation
  • Row-level security policies
  • AES-256 token encryption
  • Per-request audit logging

Performance

  • High-throughput bulk ingestion
  • Low-latency API responses
  • Efficient time-series compression
  • Optimized hypertable chunking
  • Indexed time-series queries

Quality

  • Comprehensive integration tests
  • Pydantic schema validation
  • OpenAPI spec generation
  • GitHub Actions CI/CD
  • Docker containerization

Architecture

  • 8 decoupled subsystems
  • Role-based access control
  • Event-driven data pipeline
  • Async request handling
  • Horizontal scaling ready
04

Technology Stack

Backend

  • FastAPI 0.100+
  • Python 3.11
  • Pydantic v2
  • Auth0 OAuth2

Database

  • TimescaleDB 2.23+
  • PostgreSQL 17
  • pgvector 0.5+
  • pgAdmin 4

Frontend

  • React 19
  • Vite 6
  • ECharts 5.5+
  • Tailwind CSS 3

Infrastructure

  • Docker 24
  • Raspberry Pi 5 8GB
  • NVMe SSD 2TB
  • GitHub Actions CI/CD
05

Live Environments

Explore the running applications

06

The Team

Four engineers building full-stack healthcare infrastructure.

KB

Kian Bobrow

MM

Mason Morse

GM

Grant McNatt

RR

Ryan Rudin