AI Application Developer

Write your first Python program and print to the screen.

Numbers, strings, booleans — the building blocks.

Arithmetic, powers, the math module — Python as a calculator.

Slicing, methods, and f-strings — text manipulation done right.

Make decisions and repeat yourself — but only on purpose.

Loop until a condition fails. Bail out or skip with break/continue.

Collections of things, and how to walk through them.

Carve up lists and strings with [start:stop:step].

Two more collections: tuples are fixed, sets are unique.

Key → value pairs. The most important data structure in Python.

Empty things are False. `is` vs `==`. The infamous None.

The Pythonic ways to iterate — count, pair, sequence.

Bundle up logic so you can reuse it.

Lists of dicts, dicts of lists — modelling real-world data.

Filter and transform in one line — the Pythonic way.

Catch what could go wrong, recover gracefully.

Bundle state and behavior — the basics of OOP in Python.

Parse and produce JSON — the universal data format.

GET, POST, headers, status codes, JSON — how Python talks to the web.

Concurrent code without threads. The asyncio model.

Vectorized math in Python. Auto-installs NumPy in your browser on first run.

Turn text into a numeric vector, then measure how close two pieces of text are with cosine similarity.

Store documents as (id, vector) pairs and retrieve the top-k most similar to a query — the core of every RAG retriever.

Split a long document into overlapping chunks by characters or words so each piece is small enough to embed and retrieve.

Wire embeddings, retrieval and prompt-building into one flow, then see exactly where the LLM call slots in.

Build reusable prompt templates, structure system vs user messages, and add few-shot examples — all as plain Python.

Measure how good your retriever is with precision@k and recall@k, given the ids it returned versus the ids that are actually relevant.

Typed request/response models, auto-validation, auto-docs — modern Python APIs.