ai-infra-skill

What problem does it solve?

Infrastructure for AI is complex to deploy and maintain; this skill provides a blueprint to run LiteLLM, Langfuse, RAG pipelines, and vector stores with observability in a homelab or production environment.

Core Features & Use Cases

• LiteLLM proxying for unified access to multiple providers and models
• Langfuse-based observability, tracing, and prompts management
• RAG indexing and vector search with pgvector
• End-to-end deployment, evaluation, and maintenance workflows
• Use Case: Deploy an AI assistant that mixes local and cloud models with end-to-end monitoring and cost awareness

Quick Start

Set up LiteLLM, Langfuse, and pgvector on the target VMs and run a basic health check.