Self-Hosting Overview
PolitikTok is designed to be fully self-hosted. Your data never leaves your infrastructure — all AI processing happens locally through Ollama or any OpenAI-compatible API.
Deployment Options
| Method | Best For | Complexity |
|---|---|---|
| Docker Compose | Small to medium deployments | Low |
| Manual | Full control, custom setups | Medium |
| Kubernetes | Large-scale, multi-node | High |
Architecture Overview
A production PolitikTok deployment consists of:
┌─────────────────────────────────────────────┐
│ Reverse Proxy │
│ (Caddy / Nginx / Traefik) │
│ SSL Termination │
├─────────────────────────────────────────────┤
│ PolitikTok App │
│ (Dioxus SSR + Axum API) │
├──────────┬──────────┬──────────┬────────────┤
│ PostgreSQL│ Qdrant │ Keycloak │ Ollama │
│ DB │ Vectors │ Auth │ LLM │
└──────────┴──────────┴──────────┴────────────┘Minimum Server Requirements
- CPU: 4 cores (8+ recommended for LLM inference)
- RAM: 8 GB (16+ GB recommended)
- Disk: 40 GB SSD (100+ GB if hosting LLM models)
- OS: Ubuntu 22.04 LTS or newer
Security Considerations
- All services should run behind a reverse proxy with SSL
- Keycloak handles authentication — never expose it directly
- Database ports should not be publicly accessible
- Use strong secrets for JWT, database passwords, and API keys
- Enable firewall rules to restrict access
Next Steps
- Server Setup — Prepare your server
- Docker Compose Production — Deploy with Docker
- Domain & DNS — Configure your domain
- Reverse Proxy & SSL — Set up HTTPS
- Backups — Protect your data
- Maintenance — Keep things running