AI Internals

Theory Track · diagram-first

AI Internals & Intuition

How AI systems really work — from Unicode and tokenizers through transformer attention, embeddings and RAG, training and evaluation, inference engines and GPU memory math, agents, operations, security, and enterprise architecture. 183 diagram-first lessons across 4 tracks and 12 phases, with worked numbers and diagnostics.

Companion course: Build Track — AI Project Lab ↗ walks you through implementing one of these systems end to end on a single VM. Understand the mechanism here; build it there.

Sign in

Use your email account.