Learn the whole ML system.
All material is unlocked. Follow the core dependency path, accelerate through familiar lessons, or use a role track on the roadmap.
Foundations
3 connected courses
Engineering foundations
A churn notebook cannot be reproduced by the teammate expected to ship it.
7 lessons · 1 production assignment ↗Problem framing, statistics & evaluation
An ETA model improves RMSE while increasing missed-delivery cost.
6 lessons · 1 production assignment ↗Data & feature foundations
Fraud labels arrive late and online features disagree with training data.
5 lessons · 1 production assignment ↗Classical ML
5 connected courses
Linear, probabilistic, neighbor & margin models
Support tickets need explainable routing under a 15 ms latency budget.
5 lessons · 1 production assignment ↗Trees, forests & gradient boosting
A marketplace must catch rare fraud without blocking trustworthy sellers.
5 lessons · 1 production assignment ↗Unsupervised learning, anomalies & retrieval
Novel equipment failures appear before reliable labels exist.
5 lessons · 1 production assignment ↗Forecasting & temporal ML
Regional demand shifts through promotions, holidays, and drift.
5 lessons · 1 production assignment ↗Recommendation, search & ranking
Marketplace ranking learns from biased clicks and reinforces its own mistakes.
5 lessons · 1 production assignment ↗Deep learning
3 connected courses
Neural networks from scratch
A nonlinear sensor classifier defeats every classical baseline.
7 lessons · 1 production assignment ↗Image ML
A factory has few positive examples of costly visual defects.
5 lessons · 1 production assignment ↗Sequence ML: RNNs, GRUs & LSTMs
Long support conversations must be routed while preserving temporal context.
5 lessons · 1 production assignment ↗Training systems
2 connected courses
Transformers
3 connected courses
Build a transformer from scratch
Domain autocomplete needs a model whose every component the team can explain.
5 lessons · 1 production assignment ↗Foundation-model lifecycle
A private support assistant must answer from evidence and resist manipulation.
5 lessons · 1 production assignment ↗Frontier architectures & efficient inference
Long-context generation breaks latency and cost targets.
5 lessons · 1 production assignment ↗Production
4 connected courses
Production data & ML platforms
A late feature silently corrupts every downstream prediction.
5 lessons · 1 production assignment ↗Serving & inference systems
Launch traffic causes p99 spikes and leaves expensive GPUs idle.
5 lessons · 1 production assignment ↗Reliability, monitoring & experimentation
Model quality decays while every infrastructure dashboard stays green.
5 lessons · 1 production assignment ↗Governance, fairness, privacy & security
A high-performing system creates subgroup harm and leaks sensitive information.
5 lessons · 1 production assignment ↗