Clinical AI Assistant
Multimodal clinical assistant for real-time intake, transcription, and AI documentation.
- TypeScript
- Node.js
- Azure OpenAI
- Whisper
- LiveKit
I built a voice-enabled clinical assistant that captures patient intake, streams audio in real time, and generates structured documentation using AI. The system was designed for HIPAA-safe handling, low-latency transcription, and seamless integration with clinical workflows.
What it does
The assistant listens to conversations, transcribes audio with Whisper-powered speech recognition, and uses conversational AI to generate SOAP notes, progress summaries, and session write-ups. It also supports live clinician interaction through a real-time streaming interface.
My role
I led the full-stack implementation of the assistant: building the backend services, orchestrating transcription and inference pipelines, and integrating LiveKit for real-time audio sessions. I optimized the RAG workflow for accurate clinical context and ensured the overall solution met healthcare security and compliance requirements.