Python for Java Developers → AI Engineer

Learn Python the way a Java developer thinks: 16 hands-on lessons from Hello World to FastAPI, pandas, and building an AI assistant with Claude & RAG.

Who this is for

You already know how to program: Java, Spring, JUnit, Maven. You don't need another "what is a variable" tutorial. You need to know what's different, what the Python equivalent of X is, and which libraries people actually use. Every lesson maps Python onto concepts you already know:

You know (Java)	You'll learn (Python)
record, Lombok @Data	@dataclass, pydantic models
stream().filter().map().collect()	comprehensions, generators, itertools
interfaces	Protocol (structural typing)
JUnit, Mockito	pytest fixtures, parametrize, Mock
Spring @RestController + JPA	FastAPI + SQLAlchemy
CompletableFuture, virtual threads	asyncio, async/await
Maven + pom.xml	uv + pyproject.toml

What you'll build

One project grows with you, lesson by lesson: a personal notes assistant. It starts as a script, then becomes a CLI, a REST API, a data pipeline, and finally an AI assistant that answers questions about your notes using retrieval-augmented generation (RAG).

When you finish the capstone, it works like this:

$ notes add "RAG" -b "Retrieve relevant notes with embeddings, then prompt Claude" -t ai
Added note 1: RAG
$ notes search "embeddings"
1: RAG
$ notes ask "How does RAG use my notes?"
<Claude's answer, grounded in your notes, citing [1]>
Sources: 1

Roadmap

#	Lesson	Python concepts	Libraries	Exercise
01	Setup & Hello World	interpreter, REPL, dynamic typing, f-strings, __main__	—	warm-up functions
02	Collections & comprehensions	list/dict/set/tuple, slicing, comprehensions, unpacking	collections	in-memory note store
03	Functions	defaults, *args/**kwargs, type hints, closures	—	query predicates & combinators
04	Classes the Pythonic way	@dataclass, dunders, properties, Protocols, Enum	dataclasses	Note model + NoteStore
05	Modules & packages	imports, __init__, __main__, pyproject.toml	uv	package it as notes_lite
06	Errors, files, context managers	exceptions, with, pathlib, JSON, CSV	pathlib, json, csv	persistence + atomic writes
07	Iterators, generators, decorators	yield, itertools, functools, @contextmanager	itertools	streaming + @timed/@retry
08	Testing	fixtures, parametrize, mocks, monkeypatch	pytest	write tests, find the bug 🐛
09	Command-line apps	type-hint-driven CLIs, CliRunner	typer, rich	the notes CLI
10	Validation & HTTP	models, validators, MockTransport	pydantic, httpx	typed API client
11	REST API + database	Depends, ORM, dependency overrides	fastapi, sqlalchemy	notes REST API
12	Concurrency	async/await, gather, Semaphore, Queue, GIL	asyncio	concurrent fetching
13	Data essentials	vectorization, DataFrames, plotting	numpy, pandas, matplotlib	note analytics
14	LLMs with Claude	messages, streaming, structured output, tool use	anthropic	summaries, chat, agent loop
15	Embeddings & RAG	TF-IDF, cosine search, grounding, recall@k	numpy + anthropic	"ask your notes"
16	Code quality & shipping	lint, types, logging, packaging	ruff, mypy	clean-up + capstone

Side lessons

Tools that aren't part of the main track but that you'll use throughout.

Lesson	What
Jupyter	interactive notebooks for exploring data and AI

Getting started

Requires uv (it installs Python 3.13 for you):

curl -LsSf https://astral.sh/uv/install.sh | sh
git clone https://github.com/rezaarshad/python-for-java-developers.git && cd python-for-java-developers
uv sync                                      # creates .venv with Python 3.13 and every library
uv run python lessons/01_hello/hello.py

How each lesson works

Each lesson lives in lessons/NN_name/ with:

• lesson.md: the concepts, with Java side-by-side comparisons
• demo *.py files: runnable examples (read them top to bottom)
• an exercise with TODOs:
  • Lessons 01–07: exercise.py, with assert checks in main() → uv run python lessons/NN_.../exercise.py until it prints All checks passed ✅
  • Lessons 08–16: a module plus provided pytest tests → uv run pytest lessons/NN_... -v until everything is green

Every exercise has been checked against a reference solution (not included), so if a test fails, the bug is in the exercise code, not the test.

uv run pytest lessons/09_cli -v        # test one lesson
uv run pytest                          # everything (unsolved exercises fail; that's expected)

Lesson 14 onward can call the real Claude API: set ANTHROPIC_API_KEY to run the demos. The exercise tests use fake clients, so they need no key and cost nothing.

License

MIT: use it, fork it, teach with it.