Case study
Vital Sense (MATERA) — AI clinical decision support
Sr. Full Stack Developer / Architect · AIM (Canadian IT Firm) · 2025–present
An AI clinical decision-support platform for maternal health, built on a multi-agent LangGraph pipeline with Epic FHIR integration and real-time clinician escalation.
▸ Multi-agent AI pipeline grounded in ACOG/ADA clinical guidelines
▸ Bi-directional Epic EHR integration via SMART on FHIR R4, extended to Cerner, Accuro and TELUS PS Suite
▸ Real-time clinician escalation with SignalR alerts, Mattermost and Twilio SMS fallback
▸ 8 bounded modules with enforced architecture tests and per-module Azure SQL databases
Problem
Maternal-health risk signals arrive from many systems and are easy to miss; clinicians need timely, guideline-grounded escalation rather than raw data — with strict healthcare interoperability and auditability requirements.
Approach
Architected a .NET 10 modular monolith across 8 bounded modules with enforced architecture tests, and built a hybrid .NET 10 + Python LangGraph multi-agent pipeline on Azure OpenAI. Risk Surveillance, Symptom Triage and Gestational Context agents run in parallel and are fused by a Clinical Supervisor evaluated against ACOG/ADA guidelines. Epic EHR is integrated via SMART on FHIR R4 with bi-directional write-back, and processing is event-driven on Azure Service Bus with Redis-backed deduplication and caching.