Generative AI Intermediate Launching soon

Claude Code

Agentic coding with Claude Code — driving a terminal agent through real repositories, reviewing what it writes, and shipping work you can still explain in an interview.

Agentic coding tools changed what a first-year developer’s day looks like, and interview panels have noticed. The question is no longer whether you used AI — it’s whether you can explain the code that came out of it. This track treats Claude Code as what it is: a fast, tireless pair programmer that still needs a developer who reads diffs, writes tests and knows when the answer is wrong.

You’ll work inside real repositories rather than toy snippets, because the skills that matter — scoping a change, giving an agent the right context, catching a plausible-looking mistake in review — only show up at that size.

This track assumes you can already read the language you’re working in. If you’re still building that base, start with the Programming Languages track. For the model side of things — how LLMs actually behave, and why — the Generative AI track is the companion course.

Video lessons are in production — subscribe on YouTube to catch them as they publish.

Planned syllabus

12 lessons

1 What an agentic coding tool is — and what it isn't
2 Setup: install, authenticate, first session
3 How Claude Code reads a repository
4 Prompting for code: scope, constraints, acceptance criteria
5 CLAUDE.md — teaching the agent your project's rules
6 Reading a diff properly before you accept it
7 Tests as the contract: making the agent prove its work
8 Debugging with an agent — stack trace to root cause
9 Refactors across many files without losing the thread
10 Git workflow: branches, commits and pull requests
11 Where agents fail — and the habits that catch it
12 Capstone: ship a feature end to end and defend the diff

Video lessons for this course are being published. Get a ping when they drop:

Subscribe on YouTube Join Telegram for alerts