Course · Interactive

Advanced AI Lab

Build AI from the ground up: backpropagation by hand in NumPy, real PyTorch on a free GPU via Colab, and an embedding model running live in your browser. Three tiers, one lab.

Who it's for

Learners who finished Machine Learning and want to see what's inside PyTorch and run real models — no servers, no API keys.

What you'll learn

  • Derive backpropagation with the chain rule and train a net in pure NumPy
  • Run and fine-tune a real transformer in PyTorch on a free Colab GPU
  • Load an embedding model in the browser and build semantic search
  • Understand the three tiers of doing AI: from-scratch, framework, and on-device

Course contents

  1. 01Backpropagation from scratch (NumPy)
  2. 02Real PyTorch on a free GPU (Colab)
  3. 03Semantic search with in-browser embeddings