AI-Augmented Engineering
Welcome to AI-Augmented Engineering, a course for working software engineers, tech leads and instructors who want to use AI coding assistants - Claude Code, GitHub Copilot, Cursor and their successors - to deliver more, without lowering the bar on correctness, security or accountability.
The premise is simple. The assistant writes a lot of the code now. What it cannot do is decide what to build, judge whether the result is right, or carry the responsibility when it is not. Those three things are the job, and this course is about doing them well.
Who this course is for
- Engineers who already use an assistant and want to move from autocomplete to a repeatable delivery workflow.
- Tech leads and staff engineers who need to set policies, review AI-generated changes and keep a team's velocity honest.
- Instructors and mentors who teach these practices and need a structure that survives a classroom.
You should be comfortable with Git, one backend language, basic web development and the idea of automated tests. No machine-learning background is required.
What you will learn
| Module | Theme | You will be able to |
|---|---|---|
| 1 | The AI-Augmented Engineer | Describe what changes in the job, what does not, and the new delivery loop |
| 2 | Spec-Driven Development | Write specifications an agent can execute and a test suite can verify |
| 3 | Agentic Workflows and Coding Assistants | Choose and configure Claude Code, Copilot or Cursor; engineer the context they work in |
| 4 | Full-Stack AI Delivery | Build a RAG feature end to end and wire evaluation gates into CI/CD |
| 5 | Governance, Security and Compliance | Name the risks, write the policies and leave an audit trail |
| 6 | Teaching and Leading AI-Augmented Teams | Teach a technical concept in ten minutes and measure a team's adoption |
| 7 | Agents, Sub-Agents and AI Teams | Open the agent loop, delegate to sub-agents with contracts, and run a team of specialised agents with a human in charge |
Each module has a short overview, two or three chapters and a quiz. A glossary and a list of resources close the course.
How to use this course
Read the modules in order the first time: each one assumes the vocabulary of the previous one. Afterwards, treat it as a reference. The chapters are written so that a single one can be read in twenty to thirty minutes, and the quizzes are self-correcting - click an option to see whether it is right and why.
Everything in this course is tool-agnostic unless a chapter says otherwise. Product names appear where the differences matter, because they do matter in practice, but the underlying practices (spec first, verify always, keep the human accountable) are the same whichever assistant is in your editor next year.
A note on language
The course is written in English. French, Arabic, Spanish and Portuguese versions are in progress; until they are published, the language selector in the navigation bar shows the English text under each locale.