Skip to content

vtoolsid/nutrisync

 
 

Repository files navigation

NutriSync

NutriSync is a voice-first nutrition platform that helps patients log meals through conversation and gives dietitians structured, documentation-ready summaries.

Why this exists

Dietary recalls and meal documentation eat a large share of clinical time. When intake data is incomplete or scattered, care plans suffer. NutriSync was built to capture richer meal context through natural voice (and optional photo prompts) and turn it into consistent, clinician-usable output—so nutritionists spend less time chasing details and more time on patient care.

Tech highlights

  • Mobile app — Expo, React Native, Expo Router; LiveKit real-time voice; Supabase
  • Web portal — Next.js, React, Supabase
  • Shared logic — TypeScript package used across surfaces
  • Backend-style services — Node token server and LiveKit voice agent under mobile/
  • Marketing site — Next.js app in landingpage/ (same source as the public site below)

Traction

Supported by CMU's Swartz Center Accelerator, Microsoft for Startups, and Deepgram for Startups. Led a pilot with 6 registered dietitians and 30 patients at Case Specific Nutrition before the project was put on hold.

Project status

This repository is a portfolio archive. The product was actively developed through early 2025 and is currently on hold. Setup instructions and dependencies are best-effort only—this is not a maintained open-source product, and things may not run without adjustment.

Project Overview(https://nutrisync.health)

Repository layout

Path What it is
mobile/ Main Expo app (patients + nutritionists)
web/ Next.js web portal
shared/ Shared TypeScript modules
landingpage/ Public marketing Next.js app
mobile/token-server/ LiveKit JWT token service
mobile/livekit-agent/ Conversational voice agent worker

Setup (best-effort)

  1. Prerequisites: Node.js 20+ recommended, npm, and platform tools for Expo / iOS or Android as needed.

  2. Environment: Copy env templates and fill with your own keys (never commit secrets).

    cp .env.example .env.local
    cp mobile/.env.example mobile/.env
    cp web/.env.example web/.env.local

    See .env.example, mobile/.env.example, and web/.env.example for variable names (Supabase, LiveKit, OpenAI, etc.).

  3. Mobile app

    cd mobile
    npm install
    npm run dev
  4. Web portal

    cd web
    npm install
    npm run dev
  5. Landing page

    cd landingpage
    npm install
    npm run dev
  6. Voice stack (local): Requires LiveKit (e.g. Docker), then run mobile/token-server and mobile/livekit-agent with env vars set—see mobile/README.md and mobile/livekit-agent/README.md for pointers.

License / use

MIT

About

Conversational voice AI system for clinical nutrition care. Real-time speech-to-speech pipeline using LiveKit, Deepgram ASR, Cartesia TTS, and an LLM reasoning layer with RAG to conduct dietary recalls and extract structured clinical data.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors

Languages

  • TypeScript 86.0%
  • JavaScript 10.7%
  • PLpgSQL 2.6%
  • Other 0.7%