intermediate · 6h · 5 lessons
RAG from Scratch
Build a retrieval-augmented generation pipeline that actually retrieves the right thing — without LangChain magic boxes.
By the end of this course you will be able to:
- Build an end-to-end RAG pipeline in TypeScript without a framework
- Pick the right embedding model for your data type and budget
- Diagnose why retrieval is failing using precision/recall metrics
Lessons
LESSON 1
What RAG Actually Solves (and Doesn't)
RAG isn't memory. RAG isn't fine-tuning. It's a search engine glued to an LLM — that mental model fixes most production bugs.
14 min →
LESSON 2
Chunking Strategies
Bad chunking is the #1 reason RAG retrieval fails. Get this layer right and everything downstream gets easier.
18 min →
LESSON 3
Picking the Right Embedding Model
OpenAI's text-embedding-3-large isn't always the right pick. Here's how to choose.
22 min →
LESSON 4
Vector Databases: When to Actually Use One
For under 100K chunks, you don't need Pinecone. Here's what to use instead.
16 min →
LESSON 5
Evaluating Retrieval
If you can't measure retrieval quality, you can't improve it. Build the eval harness before the product.
18 min →