Local AI Engineering with Ollama

Who Is This Book For?

This book is for people who want local AI to be something they build with, not just read about. If you can run a command and edit a file, you are qualified. The roles below will each get something different out of it.

Backend and application developers: You will learn to run a model on your own machine, talk to it over the HTTP API, and wire it into real software, moving from your first ollama run to a chat application built across nine passes: history, streaming, summarization, caching, long-term memory, function calling, and external tools. This is for the developer who wants to ship an AI feature without handing every request to a vendor.

Indie hackers and solo founders: You will learn to build and run AI products on hardware you already own, with no per-token bill eating your margin. The chapters on the SDK, custom models, and the chat application give you the spine of a product you can put in front of users. This is for the builder who wants to validate an AI idea cheaply and keep control of the unit economics.

Platform and DevOps engineers: You will learn how Ollama actually behaves under load: how long models stay in memory, how to control keep-alive, how concurrency and the request queue work, and how to inspect and manage what is loaded. This is for the engineer who has to keep a local model serving traffic and needs to predict its memory and latency before it surprises them in production.

Data and ML-curious engineers: You will learn the concepts that change your decisions (tokens, weights, quantization, the KV cache, inference parameters) and then put them to work by fine-tuning a model with LoRA/QLoRA to turn plain English into SQL. This is for the engineer who wants to teach a model a specific job without a research background or a GPU cluster.