Step-by-step mathematics behind the Attention Mechanism — the heart of ChatGPT, Gemini, and Claude. Covers word embeddings, Q·Kᵀ dot products, softmax from scratch, and final output Z = A·V with full hand calculations.

Step-by-step numeric walkthrough of gradient descent — the algorithm that trains every neural network. Covers loss functions, derivatives, the update rule, learning rate effects, all 3 types of GD, and Python code.