docs/DEPLOY_RENDER.md

Deploying Anvaya Speech AI to Render

This guide provides step-by-step instructions to deploy the Anvaya Speech Pathology & Articulation Diagnostics application on Render.

---

Deployment Architecture

The application is containerized using a production Dockerfile optimized for Render:

• Base Image: python:3.11-slim-bookworm with system audio libraries (libsndfile1, ffmpeg).
• CPU PyTorch Optimization: Uses CPU-only PyTorch wheels to reduce the image size from ~4 GB to ~600 MB, fitting easily within Render memory limits.
• Model Pre-caching: Base models are pre-cached inside the container during build time so the app boots instantly on Render.
• Dynamic Port Binding: Automatically binds to Render's dynamic $PORT environment variable.

---

Method 1: 1-Click Deployment via Render Blueprint (Recommended)

Render provides an Infrastructure-as-Code Blueprint configured in render.yaml.

Steps:

1. Push your code to GitHub:

   git add .
   git commit -m "feat: configure production Docker and Render deployment"
   git push origin main
   

2. Open Render Dashboard:
   • Go to dashboard.render.com.
   • Click New + in the top right corner and select Blueprint.
3. Connect Your Repository:
   • Select your GitHub repository (speech-model).
   • Render will detect render.yaml automatically.
4. Deploy:
   • Click Apply. Render will build the Docker container and deploy the web service.
   • Once the build completes (approx. 2 to 3 minutes), your live URL (e.g. https://anvaya-speech-diagnostics.onrender.com) will be active.

---

Method 2: Manual Web Service Setup

If you prefer to configure the service manually on Render:

1. In Render Dashboard, click New + $\to$ Web Service.
2. Select Build and deploy from a Git repository $\to$ Connect your repository.
3. Configure the service settings:
   • Name: anvaya-speech-diagnostics
   • Region: Oregon (or your preferred region)
   • Branch: main
   • Runtime: Docker
   • Dockerfile Path: Dockerfile
   • Instance Type: Starter (Recommended: 1 GB RAM, 1 vCPU) or Free
4. Set Environment Variables under Advanced:

   Variable	Value
   STREAMLIT_SERVER_HEADLESS	true
   STREAMLIT_SERVER_ENABLE_CORS	false
   STREAMLIT_SERVER_ENABLE_XSRF_PROTECTION	false
   STREAMLIT_BROWSER_GATHER_USAGE_STATS	false

5. Click Create Web Service.

---

Health Check and Live Diagnostics

• Health Check Endpoint: /_stcore/health
• Microphone Permissions: Web Audio recording works natively over HTTPS (provided automatically by Render's free SSL certificates on *.onrender.com).