Courses

Machine Learning

27 lessons

Introduction to Machine Learning → Recurrent Neural Networks (RNNs)

Regression, classification, clustering, and the evaluation that tells you whether a fit is real.

Deep Learning

16 lessons

Introduction to Deep Learning → Transfer Learning: The Donor Floor

From linear algebra and the perceptron through CNNs, transformers, and generative models.

This course is still being built. The lessons that are ready stay open; the rest appear as they ship.

Linear Algebra

14 lessons

Vectors as Lists and Arrows → SVD Product

Two vectors become columns, then two lines, then a solve. Same numbers, new picture.

Probability

16 lessons

Leftover Must Sum to 1 → Markov Leftover and Nested Expectation

A fixed budget of 1 gets split, zoomed, mixed, compared: the probability the models actually learn from.

This course is still being built. The lessons that are ready stay open; the rest appear as they ship.

Optimization

27 lessons

Lists That Agree → Barriers Turn Walls Into Bowls

A loss is a hill and training walks downhill: gradients, steps, walls, and prices — one bowl walked from f(w) toward AdamW.

This course is still being built. The lessons that are ready stay open; the rest appear as they ship.

Python for ML

33 lessons

Python as an ML Calculator → Capstone Hardening and Reflection

Run real Python in the browser: variables, lists, numpy, pandas, and plots. Every output executed, never handwritten.

This course is still being built. The lessons that are ready stay open; the rest appear as they ship.

Harness Engineering

83 lessons

The same model, five harnesses → The lean operator

Build the machinery around the model: loops, tools, verifiers, sandboxes, and evals — in the browser where the lab allows.

This course is still being built. The lessons that are ready stay open; the rest appear as they ship.

Evals Engineering

27 lessons

Why evals decide whether AI products live → End-to-end RAG harm checks

Measure, debug, and ship reliable AI systems: error analysis, binary graders, trace diagnostics, and release gates.

This course is still being built. The lessons that are ready stay open; the rest appear as they ship.

Memory Engineering

13 lessons

The forgetful agent → Reranking, filtering & knowing when to abstain

What an agent keeps across sessions: a typed store, hybrid retrieval, an adjudicated write path, temporal truth, gates, and the harness that measures them.

This course is still being built. The lessons that are ready stay open; the rest appear as they ship.