A HANDS-ON FIELD GUIDE TO RAG
Build answers.
Show the
evidence.
Learn the engineering behind answers you can trace. Turn documents into searchable knowledge, build a retrieval pipeline, and test every claim.
A source worth
coming back to.
Students may take up to 10 days of medical leave in a semester.
KAVERI INSTITUTE · FICTIONAL EXAMPLEUp to 10 days per semester.[1]
[1] Medical Leave · Kaveri InstituteFrom the first search
to a system you trust.
A LITTLE RETRIEVAL PRACTICE
Keep the ideas within reach.
Based on your self-checks · saved in this browser · no reminders or streaks
A language model knows a lot.
It doesn’t know your documents.
Retrieval-augmented generation (RAG) gives it the missing context. Find the right passages, bring them into the conversation, and make the answer traceable. This course shows you how, one experiment at a time.
Find the signal.
Search beyond exact words. Retrieve the passages that actually matter.
Build the context.
Give the model useful evidence, with enough room for a good answer.
Check the answer.
Trace every claim to a source. Measure what works and fix what doesn’t.
02 / TRY THE IDEA
A question in.
A source out.
Three questions. A tiny campus knowledge base. Pick a question and watch a passage become a cited answer.
A local retrieval demo using fictional campus policies. No model calls.
Enter the playgroundHow many days of medical leave can I take?
The useful part isn’t just the answer.
It’s knowing where it came from.
03 / YOUR LEARNING PATH
Start small.
Build something real.
A self-paced course for people comfortable with Python. Search and machine learning concepts start from first principles.
Learn to find.
From inverted indexes to embeddings. Understand how search finds what you mean.
Explore the foundationsConnect the pieces.
Parse, chunk, embed, retrieve. Build pipelines for text, graphs, and multimodal documents.
Build your pipelineMake it hold up.
Evaluate the answers. Debug the misses. Keep your system fast, secure, and affordable.
Go beyond the prototype04 / LEARNING BY DOING
Your browser is
the laboratory.
Read a clear explanation. Move a slider. Break a ranking. Run a Python example. Every lesson pairs the idea with something you can actually try.
YOUR NEXT GOOD QUESTION STARTS HERE.
Build it.
Back it up.
Start with search. Follow the evidence.
Build a little understanding every day.
0 / 92 lessons completed · saved in this browser