# 🧠 BetelgeuseBytes AI Platform — Documentation

This documentation describes a **self-hosted, CPU-first AI platform** running on Kubernetes,
designed to power an **Islamic Hadith Scholar AI** and future AI/data projects.

## 📚 Documentation Index

- [Architecture](ARCHITECTURE.md)
- [Infrastructure](INFRASTRUCTURE.md)
- [Full Stack Overview](STACK.md)
- [Deployment & Operations](DEPLOYMENT.md)
- [Observability](OBSERVABILITY.md)
- [Roadmap & Next Steps](ROADMAP.md)
- [Future Projects & Use Cases](FUTURE-PROJECTS.md)

## 🎯 Current Focus

- Hadith sanad & matn extraction
- Narrator relationship modeling
- Knowledge graph construction
- Human-in-the-loop verification
- Explainable, sovereign AI

## 🧠 What each document gives you
### ARCHITECTURE

- Logical system architecture

- Data & control flow

- Networking and security model

- Design principles (CPU-first, sovereign, observable)

- What the architecture enables long-term

This is what you show to **architects and senior engineers.**

### INFRASTRUCTURE

- Hetzner setup (dedicated, CPU-only, SSD)

- Node roles and responsibilities

- Kubernetes topology

- Cilium networking

- Storage layout on disk

- Namespaces and isolation strategy

This is what you show to **ops / SRE / infra people.**

### STACK

- Exhaustive list of every deployed component

- Grouped by domain:

    - Core platform

    - Databases & messaging

    - Knowledge & vectors

    - ML & AI

    - Automation & DevOps

    - Observability

    - Authentication

For each: **what it does now + what it can be reused for**

This is the **master mental model** of your platform.

### DEPLOYMENT

- How the platform is deployed (kubectl + GitOps)

- Deployment order

- Operational rules

- Backup strategy

- Day-2 operations mindset

This is your ***runbook starter.***

### ROADMAP

- Clear technical phases:

    - Neo4j isnād schema

    - Authenticity scoring

    - Productization

    - Scaling (GPU, multi-project)

This keeps the project ***directionally sane.***

### FUTURE-PROJECTS

- Explicitly documents that this is **not just a Hadith stack**

- Lists realistic reuse cases:

    - RAG

    - Knowledge graphs

    - Sovereign AI

    - Digital humanities

    - Research platforms

This justifies the ***investment in infra quality.***