RAG From Scratch: build a document Q&A chatbot, end to end
A live, hands-on workshop where you build a production-style RAG chatbot from a starter repo — not a finished one — and understand why every component exists before you write a line of code.
Understand → Build → Break → Debug → Improve
Every component is introduced by answering why it exists before writing code. You're not here to memorize libraries — you're here to understand the engineering decisions behind them, so you can modify this system or build your own afterward.
A starter repo with 7 guided missions — not a finished project
Each mission is a set of TODOs, introduced with a discussion of the “why” before any implementation.
- 01
Teach the AI to read documents
Load raw files — PDFs, text, docs — and turn them into clean, usable text.
- 02
Split documents into meaningful chunks
Understand why chunk size and overlap change everything downstream.
- 03
Generate embeddings
Convert text into vectors and see what those numbers actually represent.
- 04
Store vectors in a database
Set up a vector store and understand what it's optimized for.
- 05
Retrieve relevant context
Implement similarity search and tune what counts as 'relevant.'
- 06
Connect an LLM
Wire retrieved context into a prompt and get your first real answer.
- 07
Improve answer quality
Diagnose weak answers and fix the pipeline — not just the prompt.
On purpose — so you learn to diagnose, not just trust
You'll intentionally push the system into failure, watch answers get worse, and learn to trace the failure back to its cause.
- Extremely small chunk sizes
- No overlap between chunks
- Wrong top-K values
- Poor prompts
- Missing retrieval entirely
Everything you need, nothing you don't
- Live, instructor-led online session
- Full workshop recording
- Starter repository with 7 guided missions
- Certificate of completion
- Post-workshop resource pack
Apply code FLAT100 at checkout
Continue to register2:00 PM – 6:00 PM IST · You'll get a calendar invite by email
Before you register
Do I need prior AI/ML experience?
No. You need to be comfortable writing basic code — Python or JavaScript. Every AI concept is taught from first principles.
What if I miss the live session?
You'll get the full recording, so you can go through every mission at your own pace.
What do I need installed?
Setup instructions are emailed after registration, well before the workshop, so you arrive ready to build.