How to use this guide

Part 0 (Start here): setup with uv, how this guide fits the series, a Go ↔ Rust ↔ Python phrasebook, the labs, and an honest answer to "why Python for AI?". Read the phrasebook even if you skip everything else.
Beginner (Part I): the interpreter, names and objects (the chapter that explains most Python bugs), numbers, text and bytes, control flow, functions, collections, exceptions, files, modules. Ends with a 🔎 Checkpoint and two terminal projects.
Intermediate (Part II): iterators and generators, decorators, classes, the data model (dunder methods), protocols, type hints, the standard library you will use every week, testing with pytest, and packaging. Ends with a checkpoint, a grep clone, and an LRU cache.
Advanced (Part III): how CPython runs your code, memory and garbage collection, the GIL and free-threaded Python, multiprocessing, asyncio, descriptors and metaclasses, and performance work.
Systems Python (Part IV): the OS guide in Python: syscalls, processes and signals, crash-safe files, /proc, mmap, Python as a better shell, and a Linux container with os.unshare.
Network Python (Part V): the networking and HTTPS guides in Python: sockets four ways, a DNS client from raw bytes, HTTP from a bare socket, HTTP clients done right, a production API, and TLS/mTLS with ssl.
Security-focused Python (Part VI): Python's own attack surface (pickle, eval, shell=True, archive extraction), cryptography done correctly, web security, the supply chain, and an auth service. Ties to both security guides.
Python for data and AI (Part VII): NumPy, pandas and Polars, scikit-learn, autograd from scratch, PyTorch, Hugging Face, data pipelines for training, and LLM apps (RAG, tool-calling agents, an MCP server) against a local open-weight model. This Part hands you to the AI guide.
Open source walkthrough (Part VIII): how to read a real codebase, then requests traced from requests.get() through sessions, adapters, urllib3's connection pool, TLS, and the socket. Ends with running its test suite and making a change.
Expert and professional (Part IX): library and API design, logging and observability, debugging, containers, an anti-pattern catalog, and the capstone: one request through OS → network → security → AI, in Python.
Appendices: a traceback decoder, a tooling cheat sheet, a glossary, a 90-day plan, and the other guides and books this one draws on.

Conventions:

Code targets Python 3.13+. Every complete program was run with CPython 3.13.9 (macOS) and the official python:3.14-slim image (Linux). Features that need 3.14 are marked (3.14+). Third-party package versions are the ones current when the guide was written; they are pinned in each pyproject.toml snippet.
Snippets that are wrong on purpose start with # WRONG or end with the traceback they produce. Each one is followed by the fix.
>>> lines are typed at the interactive interpreter (the REPL). The line after them is what Python prints.
Real-world example boxes connect a feature to code you will actually write. War story boxes describe a real class of incident.
Across the series notes link to the chapter in another guide that covers the same topic from the OS, network, security, Go, Rust, or AI side.