Before you build anything with AI, you should understand what is really happening inside the model. This module removes the magic: by the end, words like transformer, token and embedding are tools you think with, not buzzwords you nod along to.
Most AI courses jump straight to tools and prompts. We start one level deeper, because engineers who understand why a model behaves the way it does can debug, design and reason where others can only guess.
No prior coding experience is required. Every idea is built up from scratch, hands-on, with mentors in the room.
How the transformer architecture works and why attention lets a model weigh every word against every other word — the idea behind the entire modern AI wave.
What text becomes inside a model: how sentences are split into tokens, and how embeddings turn meaning into geometry that a machine can compute with.
How models learn from vast text corpora, what fine-tuning and RLHF actually change, and why the same base model can behave so differently after alignment.
What a context window really is, why it limits what a model can "remember" in a conversation, and how that shapes every application you will ever build.
Why the same prompt gives different answers: top-p, top-k and temperature, and how to choose settings deliberately instead of accepting defaults.
The honest truth about hallucination — where it comes from, why it cannot be fully "fixed", and the engineering patterns that keep it under control.
Module 1 is where every Bonami graduate begins. If you are ready to learn AI properly — from the inside out — we would be glad to have you.