Shared Foundations

These pages cover concepts that both the Beginner and Advanced courses rely on. They provide just enough theory to understand system behavior without becoming a separate ML course.

Why These Topics Are Shared

Both courses need a common vocabulary for:

• Tokenization and context windows — to reason about prompt size and cost
• Embeddings and similarity — to understand retrieval
• Transformers and attention — to understand why context design matters
• Prompt patterns — to build reliable outputs

Pages in This Section

Foundational Concepts

Page	What You'll Learn
LLM basics for engineers	What LLMs are, how they work, and why they're probabilistic
Transformer basics	Self-attention, context windows, and architecture intuition
Tokenization and context windows	Token budgeting, context limits, and cost reasoning

Retrieval Foundations

Page	What You'll Learn
Embeddings and similarity	Vector representations, cosine similarity, nearest-neighbor
Chunking and retrieval primitives	Document preparation, top-k retrieval, and reranking

Reference Material

Page	What You'll Learn
Prompt and output patterns cheatsheet	Reusable patterns for prompting and structured output
Glossary	Definitions of key terms used throughout both courses
Design checklists	Decision checklists for GenAI system design
Evaluation worksheet	Practical guide to building evaluations

How to Use These Pages

For the Beginner Course

Read these pages before or alongside the lessons:

1. LLM basics for engineers → Before Lesson 1
2. Tokenization and context windows → Before Lesson 2
3. Embeddings and similarity → Before Lesson 4
4. Prompt and output patterns cheatsheet → Reference throughout

For the Advanced Course

These pages serve as a refresher and common vocabulary:

• Context engineering recap → Lesson 3
• Retrieval architecture → Lesson 4
• Design checklists → Reference for architecture decisions
• Glossary → For consistent terminology

Key Principles

1. Enough Theory, Not Too Much

These pages teach concepts needed for building, not for publishing papers. We skip:

• ❌ Matrix derivations and math
• ❌ Full training pipeline details
• ❌ Exhaustive benchmark comparisons

We focus on:

• ✅ Mental models that predict system behavior
• ✅ Engineering decisions and tradeoffs
• ✅ Practical patterns that work

2. Theory → Practice → Theory

Each page includes:

1. Conceptual explanation with diagrams
2. Practical code examples
3. Design implications

3. Consistent Terminology

Both courses use these terms consistently. If you see a term you don't recognize, check the Glossary.

Next Steps

• Continue to LLM basics for engineers to start learning
• Or jump directly to a course:
  • Beginner Course: Start here
  • Advanced Course: Start here