interactive deep learning course

Deep Learning — from ∂loss/∂w
to a fine-tuned transformer.

Three chapters of just-enough math, four chapters of PyTorch (tensors, autograd, training loop, CNNs), and a capstone where you fine-tune a real transformer on a real dataset. Every lab runs in your browser.

Start the course →Open the playground

Beginner

Intermediate

Advanced

Expert

What you'll come out knowing

The math, demystified

Vectors, gradients, distributions, cross-entropy — taught the way a practitioner thinks about them, not the way a textbook proves them.

PyTorch fluently

Tensors, autograd, nn.Module, optimizers, the training loop. Every line of every production script will look familiar.

A real fine-tune, end-to-end

Take a pretrained transformer, fine-tune it on a real dataset, and ship it. The exact path of every modern AI startup.

Deep Learning — from ∂loss/∂w to a fine-tuned transformer.

What you'll come out knowing

Deep Learning — from ∂loss/∂w
to a fine-tuned transformer.