MLCopilot Platform — Production Deployment Guide

This guide provides step-by-step instructions for deploying MLCopilot to production environments using Docker Compose, container registries, or Kubernetes.

1. System Requirements

Hardware Requirements

CPU: 4 cores minimum (8 cores recommended for local Ollama inference)
RAM: 16 GB minimum (32 GB recommended if running llama3.1:8b locally)
Disk: 50 GB NVMe/SSD storage for database, vector indexes, and object storage

Software Dependencies

Docker Engine: v24.0+
Docker Compose: v2.20+
Ollama: v0.3.0+ (Installed locally or on an inference server)

2. Pre-Deployment Configuration

Clone the Repository:

git clone https://github.com/your-org/MLCopilot-Platform.git
cd MLCopilot-Platform

Setup Production Environment Variables: Copy the example production environment file:
```
cp production.env.example .env
```
Generate Secure Secrets: Generate a 64-character JWT secret key:
```
openssl rand -hex 32
```
Update JWT_SECRET and POSTGRES_PASSWORD in .env.

3. Production Deployment with Docker Compose

Run the entire platform stack in production mode:

docker compose -f docker-compose.yml up -d --build

Stack Components:

web: Next.js Frontend (Port 3000)
api: FastAPI Backend (Port 8000)
postgres: PostgreSQL 16 + pgvector extension (Port 5432)
redis: Redis 7 Cache & Task Broker (Port 6379)
minio: Object Storage for document uploads (Ports 9000, 9001)

4. Verification & Health Probes

Verify that all service health checks pass:

Liveness Probe:

curl -s http://localhost:8000/api/v1/health/live

Expected Output: {"status":"ok"}

Readiness Probe:
```
curl -s http://localhost:8000/api/v1/health/ready
```
Expected Output: {"status":"ok","checks":{"database":"ok","redis":"ok","storage":"ok"}}
LLM Provider Probe:
```
curl -s http://localhost:8000/api/v1/health/llm
```
Expected Output: {"provider":"ollama","model":"llama3.1:8b","display_name":"Ollama • llama3.1:8b","status":"healthy"}

5. SSL & Nginx Reverse Proxy Setup

Recommended Nginx configuration for SSL termination and SSE streaming support:

server {
    listen 443 ssl http2;
    server_name mlcopilot.yourdomain.com;

    ssl_certificate /etc/letsencrypt/live/mlcopilot.yourdomain.com/fullchain.pem;
    ssl_certificate_key /etc/letsencrypt/live/mlcopilot.yourdomain.com/privkey.pem;

    # Frontend Next.js Proxy
    location / {
        proxy_pass http://localhost:3000;
        proxy_set_header Host $host;
        proxy_set_header X-Real-IP $remote_addr;
    }

    # API & SSE Stream Proxy
    location /api/ {
        proxy_pass http://localhost:8000;
        proxy_set_header Host $host;
        proxy_set_header X-Real-IP $remote_addr;
        
        # SSE Streaming headers
        proxy_buffering off;
        proxy_cache off;
        proxy_set_header Connection '';
        proxy_http_version 1.1;
        chunked_transfer_encoding on;
    }
}

6. Maintenance & Backups

Database Backup:

docker compose exec postgres pg_dump -U mlcopilot mlcopilot_db > backup_$(date +%F).sql

Logs Inspection:
```
docker compose logs -f api
```

FilesExpand file tree

DEPLOYMENT.md

Latest commit

History