Build an AI platform you understand, secure, and control with The Self-Hosted AI Builder's Guide. This practical book shows developers, system administrators, homelab users, and small teams how to deploy and operate self-hosted AI services using Docker, local models, and open source components. Learn how Ollama, LiteLLM, and AnythingLLM work together for local model serving, routing, and private chat. Build document-intelligence and RAG pipelines with Docling, embeddings, reranking, and pgvector. Add batch and real-time transcription with Whisper and WhisperLive, generate speech with Kokoro, connect authenticated MCP tools, and optionally run a containerized goose agent. The book emphasizes deployment and day-to-day operation rather than abstract AI theory. Its Linux examples provide CPU-only starting points and NVIDIA CUDA acceleration where supported. You will also learn how to choose an appropriate architecture, understand the data boundaries of local models and optional hosted providers, secure exposed services, troubleshoot failures, protect persistent data, create backups, restore services, and plan upgrades.
Companion open source repositories provide maintained deployment files and commands. The book supplies the architecture, integration guidance, security practices, and operating knowledge needed to use those resources as one coherent, maintainable AI platform.