---
title: AI Applications Pipeline
created: 2026-04-28
updated: 2026-04-28
type: concept
tags: [concept, ai, services]
sources: [../../homelab/architecture.md]
---

# AI Applications Pipeline

Local AI/ML stack running on ubuntu with GPU acceleration (GTX 1080 8GB), plus AI-powered applications that use LLM inference.

## Core AI Infrastructure

| Service | URL | Purpose |
|---------|-----|---------|
| Ollama | localhost:11434 | Local LLM inference (GPU via GTX 1080) |
| Qdrant | ubuntu:6333 | Vector database for OpenCode cluster memory |
| Faster Whisper Server | — | Speech-to-text (Whisper) |

## AI Applications (7 containers)

| Application | Description |
|-------------|-------------|
| AI Job Pipeline (backend + frontend) | AI task orchestration |
| AI Alert Aggregator (backend + frontend + postgres) | Alert intelligence |
| AI Media Intelligence (backend) | Media analysis |
| AI Subscriptions | Subscription management |
| Homelab Inventory (backend) | Infrastructure inventory |

## Immich ML

| Component | Description |
|-----------|-------------|
| Immich Server | Photo/video management |
| Immich ML | Machine learning on GPU |
| Immich Postgres | Dedicated PostgreSQL (pgvecto-rs extension) |
| Immich Redis | Caching |

## OpenCode Embeddings

OpenCode instances across the cluster use:
- **Ollama** — generating embeddings for vector memory
- **Qdrant** — storing shared vector memory across OpenCode cluster

## Related

- [[opencode-cluster]] — OpenCode cluster using this AI infrastructure
- [[ubuntu]] — Hosts GPU (GTX 1080) and all AI services
- [[jellyfin]] — Media server with AI features
- [[../../homelab/docs/ai-applications.md]] — AI applications documentation