Clinical Voice AI
Voice-driven medical receptionist & triage operations platform.
Engineered a real-time, asynchronous medical operations system and token authentication engine using Python, aiohttp, and PostgreSQL 16 to manage appointment workflows and issue WebRTC JWT tokens.Architected a highly available dual-database infrastructure leveraging asyncpg connection pooling for PostgreSQL, complete with automated migrations and an embedded SQLite failover mechanism to ensure continuous uptime during network disruptions.Strengthened infrastructure protection by configuring an Nginx ingress proxy inside containers featuring TLS 1.3 encryption, CORS enforcement, and rate-limiting rules.Containerized and deployed distributed cloud microservices with Docker and Kubernetes, seamlessly coordinating API endpoints, background media workers, datastores, and web frontends.
Real-Time Telephony & Resilient Medical Infrastructure
Medical receptionist workflows demand uninterrupted uptime and sub-second voice latency. By pairing LiveKit WebRTC with OpenAI GPT-4o and custom aiohttp service workers, the platform delivers fluid natural conversations while automating patient scheduling.
Dual-Database Failover
To prevent missed patient records during network or database maintenance, an embedded SQLite failover buffer queues incoming events locally and synchronizes back to PostgreSQL once connectivity restores.