45 courses · The first chapter of every course is free to read
Go from zero to a working movie recommender in one beginner-friendly course.
Understand ML from scratch and build simple models that find patterns.
Train a simple model and clearly explain why it makes each choice.
Learn to make simple predictions with machine learning—no coding needed.
Turn raw model scores into decision-ready probabilities you can trust.
Build ranking models that improve NDCG—from pairwise loss to LambdaMART.
Train, tune, and calibrate tree models you can trust in production.
Turn skewed labels into reliable decisions with costs, thresholds, and calibration.
Build Bayesian models that predict with calibrated uncertainty.
Ship ML with confidence using drift detection, checks, and actionable alerts.
Make product decisions with causal proof—even when A/B tests fail.
Build gradient descent from scratch and make it converge on real data.
Tune smarter: Bayesian search + multi-fidelity + early stopping, at scale.
Cluster, segment, and detect anomalies with real-world unsupervised workflows.
Ship fraud and churn models by tuning precision, recall, and thresholds.
Turn user-item graphs into ranked recommendations with GNN-ready pipelines.
Engineer better tabular features without leakage—validated and production-ready.
Build, validate, and ship forecasting models that hold up in backtests.
A practical scikit-learn workflow for picking the right model with confidence.
Derive core ML math, then implement it cleanly and correctly in code.
From data to deployed models: learn ML with Python the modern way.