Production Infrastructure Guide

VPS & Production Server Deployment

FineTuneMyAI uses a decoupled architecture: your VPS serves as the sovereign control plane, while your local Apple Silicon Macs or Linux RTX workstations act as zero-data-leak compute nodes.

Control Plane vs. Compute Sovereignty

You do not need expensive cloud GPUs on your VPS. The VPS runs the Next.js control plane, model registry metadata, and MariaDB database. Compute-heavy LoRA/QLoRA training and high-throughput inference run on your paired physical workstations via the secure pairing protocol.

1. Server Specifications

Minimum CPU

1 vCPU (x86/ARM)

2+ vCPUs recommended

Memory

2 GB RAM + 2 GB Swap

Prevents build OOM

Operating System

Ubuntu 22.04 / 24.04

Debian 12 / Rocky 9 also supported

2. Method A: Turnkey Docker Compose (Recommended)

Fastest Setup

Docker Compose orchestrates MariaDB 11.4 and the Next.js production web container with automated database initialization, migration, and health checks.

# 1. Install Docker Engine if not installed

curl -fsSL https://get.docker.com | sh

# 2. Clone repository

git clone https://github.com/kyrondataTech/fineTuneMyAi.git /var/www/finetunemyai

cd /var/www/finetunemyai/gemini

# 3. Launch isolated production containers

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

3. Method B: Native Systemd Service

Run natively on the host using Node.js 20 LTS and systemd for minimal overhead and automatic background recovery.

# 1. Install Node.js 20 LTS and build application

npm ci --prefer-offline

npm run build

npm run db:setup

# 2. Enable & start systemd service

sudo cp scripts/vps/finetunemyai.service /etc/systemd/system/

sudo systemctl daemon-reload && sudo systemctl enable --now finetunemyai

4. Method C: Automated Deployment Script

Our interactive deployer checks system resources, configures swap to prevent OOM errors, generates cryptographically secure JWT secrets, and runs health verification.

bash scripts/vps/deploy.sh

5. Nginx Reverse Proxy & Free SSL

FineTuneMyAI includes pre-configured Nginx reverse proxy templates supporting Server-Sent Events (SSE) streaming for real-time inference and training progress, plus 500MB payload support for dataset uploads.

# 1. Install Nginx and Certbot

sudo apt install -y nginx certbot python3-certbot-nginx

# 2. Install site configuration

sudo cp vps/nginx/finetunemyai.conf /etc/nginx/sites-available/

sudo ln -s /etc/nginx/sites-available/finetunemyai.conf /etc/nginx/sites-enabled/

sudo nginx -t && sudo systemctl reload nginx

# 3. Request free Let's Encrypt SSL certificate

sudo certbot --nginx -d your-domain.com

6. Verification & Health Monitoring

The control plane exposes a lightweight health endpoint at /api/health compatible with Uptime Kuma, AWS Target Groups, and Docker healthchecks:

$ curl http://127.0.0.1:3050/api/health

{"status":"healthy","service":"finetunemyai-control-plane","database":{"engine":"mariadb","status":"connected"}}

← Daemon Installation Guide Pair Worker Workstations