AI-Powered Development Course

1

Fundamentals

Week 1 covers how LLMs actually work: tokenization, context windows, log probabilities, and why the same prompt gives different results at different temperatures. Understanding these mechanics changes how you write prompts.

2

Developer Tools

Cursor, Claude Code, and Kilo side by side - what each one actually does well, where each breaks down, and when you'd pick one over the other. The goal is making an informed choice, not picking the tool with the best marketing.

3

Integration Workshop

You build an MCP server from scratch that connects Claude Code to an external API. The same week covers Claude Code hooks and agent skills - the parts of the toolchain most developers skip because they look optional until they've wasted enough time without them.

4

Development Methodologies

Week 3 probably went slower than expected. This week explains why. Spec-driven development, vibe coding, and BMAD aren't competing philosophies - they're different tools for different situations, and knowing which to use stops a lot of debugging loops before they start.

5

Production Integration

What changes when AI-assisted development moves from one person's laptop to a team's shared workflow. Cost tracking, LLM observability, security policy, and local model deployment - the parts nobody explains until something breaks or the API bill arrives.

6

Bonus Materials

Bonus material for the situations the main curriculum doesn't have room for: working in large codebases, team-wide AI configurations, visual workflow tools like cc-wf-studio and CCW, and coding on mobile. Optional, but worth reading if any of those are a current problem.