You have used ChatGPT and Claude. You keep hearing Hugging Face, Ollama, LM Studio, GGUF, quantization — and it sounds like a world built for developers that you are not supposed to touch. It isn’t. Local AI is four pieces, and once you can see them, the whole field stops being a wall of jargon and starts being a set of decisions you are perfectly equipped to make.
By the end of this course you will have an open model running on your own laptop, doing real work on your own private files, with the network switched off. Not a demo — a workflow that reads a folder and writes a document somebody can actually use.
What makes this course different: nothing here is faked.
Every lab is built from a genuine captured session on a real machine — the real pages, the real application windows, and the real unedited command output. When a model gets something wrong, that stays in — because what you do next is the actual skill. In Section 6 you will watch a real evaluation where a small model catches every planted problem and raises sixteen false alarms; then watch a prompt fix eliminate the false alarms and destroy the catch rate; then watch the model that finally worked turn out to be the one nobody expected. You cannot learn that from a scripted success.
