About This Curriculum
Large language models have fundamentally changed what's possible in software development. But using them effectively requires more than knowing which button to click—it requires mental models for how they work, what they're good at, and where they fail.
This curriculum builds your intuition for AI-assisted development. You'll learn to craft effective prompts, evaluate AI-generated code, understand security considerations, and develop workflows that leverage AI capabilities while maintaining human oversight. The goal isn't to replace your thinking—it's to amplify it.
Two curricula, one goal. LLM Fundamentals pairs with the Software Engineering (SWE) Curriculum. AI can write code fast, but only humans can make it good. Understanding both how AI works and how software engineering works is essential for the future of development.
Curriculum Structure
The curriculum spans four tiers of progressive depth. Tier 1 builds foundational mental models during bootcamp week. Tiers 2-4 expand into advanced techniques, professional workflows, and broader implications—introduced as just-in-time learning during project work.
Tier 1: Mental Models
How AI actually works
- What is machine intelligence?
- How LLMs work
- The capability inflection point
- Capabilities & limitations
- Your first conversation
Tier 2: Core Skills
Effective AI interaction
- Reading AI-generated code
- Prompting fundamentals
- Multi-modal models
- Reasoning models
- Context & memory
Tier 3: Advanced Practice
Professional workflows
- Prompt injection & security
- Agentic loops
- Developer workflows
- Human-AI partnership
- Iterating with AI
Tier 4: Broader Context
Ethics and implications
- Bias & fairness
- Privacy & data
- The future of work
Learning Materials
Each topic is available in two formats. The slide deck provides structured presentations for guided instruction and discussion. The interactive book offers deeper exploration with hands-on exercises, self-check questions, and concept reviews.
Key Themes
AI is a tool, not a replacement
Language models are remarkably capable—and remarkably limited. They can generate code faster than you can type, but they can't understand your goals, validate their own output, or take responsibility for the result. Effective AI use means knowing when to trust, when to verify, and when to step back.
Prompting is thinking out loud
Good prompts aren't magic incantations. They're clear expressions of what you want, what context matters, and what constraints apply. Learning to prompt well improves your ability to articulate problems—a skill that transfers far beyond AI interaction.
Security is everyone's job
AI introduces new attack surfaces. Prompt injection can manipulate model behavior. Generated code may contain vulnerabilities. Understanding these risks isn't optional—it's part of being a responsible developer in the AI era.
The future is collaborative
AI won't replace developers—but developers who use AI effectively will have significant advantages. The goal is human-AI collaboration that combines AI speed with human judgment, creativity, and accountability.