Lead projects built by Claude Code, and understand every decision.
An AI coding agent can build almost anything you describe, including the wrong thing, confidently. This path gives you everything an engineer knows, so you can lead the project, direct the agent and catch mistakes before your users do.
Who this path is for
- You already work as a software engineer, but you never studied it.
- You're about to lead big projects: several teams, months of work, real users.
- Claude Code writes the code. You decide, plan, check and ship.
What you'll be able to do
- 1Explain any system end to end, from a click to the database and back.
- 2Make and defend technical decisions, with their costs and risks, in writing.
- 3Turn a business goal into a plan an AI agent can execute.
- 4Direct AI coding agents safely, even several at once.
- 5Read a diff and a test well enough to accept, reject or ask for a change.
- 6Ship and run it: tests, security, deploys, monitoring, incidents, costs.
- 7Lead people: explain trade-offs, manage risk, know when to call an expert.
Your path
9 stages, in order. Start by learning how to direct an AI coder, then learn how every part of a system works, then ship and lead. Every module of the course is in here, once.
- 1.Direct an AI coder0% mastery
- 4.Run in production0% mastery
- 5.Safe, tested, affordable0% mastery
- 6.Build AI features0% mastery
- AI FoundationsStep 6.1 · 0/5 lessons
- LLM EngineeringStep 6.2 · 0/3 lessons
- Embeddings & Vector DatabasesStep 6.3 · 0/3 lessons
- RAGStep 6.4 · 0/4 lessons
- AI AgentsStep 6.5 · 0/4 lessons
- Tool CallingStep 6.6 · 0/3 lessons
- MCPStep 6.7 · 0/2 lessons
- AI MemoryStep 6.8 · 0/2 lessons
- AI EvaluationsStep 6.9 · 0/3 lessons
- 8.Ship with Claude Code0% mastery
The four rules
You own the decisions
What to build, what data to keep, what can fail, what it may cost. Claude Code types; it never decides these silently.
Ask for a plan first
For anything risky, have Claude explain its plan before it touches a file. Approving a plan you understand is the real job.
Verify, don't trust
Run it, read the diff, check the tests, look at the numbers. "Done" from an agent is a claim, not a fact.
Know the red flags
Secrets in code, disabled checks, invented libraries, personal data cached. Every lesson lists the ones for its topic.
How to use each lesson
- 1Take the skip check at the top of each lesson. If you already know it, jump to the practice.
- 2Read the episode and guess first. The story is the problem you'll have to solve with Claude.
- 3Read the explanations to understand, not to type. You must know what the code does and why, not write it from memory.
- 4Do the For vibe coders section: the decisions you own, an instruction for Claude Code, what to check, red flags, and a review drill.
- 5Never skip step 4 of the practice ladder (Decide). Writing the decision is exactly what you'll do at work.
- 6After each module, do the On the job mission on your own real project, and tick the "can you do this now?" list.