Learning paths
Each path is a short, ordered sequence of articles and labs. Follow one from the top; every step says why it comes where it does.
How language models read text
Start here if you use language models but want to understand what happens to your text before the model sees it, and why limits and costs behave the way they do.
You will be able to
- Explain what a token is and estimate token counts
- Plan a context window budget for a real application
- Explain why long inputs can be used unreliably
Build and evaluate RAG systems
A practical path from the idea of retrieval-augmented generation to a system you can measure and defend. Finish by building a small RAG pipeline yourself.
You will be able to
- Explain each stage of a RAG pipeline
- Choose and test a chunking strategy
- Compare keyword, vector and hybrid retrieval
Machine learning methods that hold up
The habits that decide whether a model's score means anything. Useful for students writing a thesis, and for anyone who has seen a great notebook score fail in production.
You will be able to
- Split data correctly, including by time and group
- Diagnose overfitting and underfitting from scores
- Use cross-validation and read its spread
Network and web security basics
Ground truth for security work: what normal traffic looks like, and the response headers that protect users. Includes a hands-on Wireshark lab on your own traffic.
You will be able to
- Explain the differences between TCP and UDP
- Find and explain a TCP handshake in a packet capture
- Read a site's security headers and suggest improvements