RAG tutorial: Python and LlamaIndex examples, free RAG courses, and RAG analysis

A good RAG tutorial walks you through building retrieval-augmented generation end to end: load documents, split them into chunks, embed and index them, retrieve relevant chunks for a question, and have a language model answer from them with citations, then evaluate and improve. I built RAG.NextUpgrad, a production RAG platform with hybrid search and reranking, so this page gives a practical tutorial path, Python and LlamaIndex examples, GitHub projects, free courses and certificates, what RAG analysis means, and, for sewers, the rag quilt.

What is a good RAG tutorial path?

Build the simplest working pipeline first, measure it, then improve the weakest stage. Most learners go wrong by adding advanced techniques before their basic retrieval works.

Also asked as: tutorial rag

A tutorial path:

• Step 1: answer questions over one PDF with chunking, embeddings and top-5 retrieval.
• Step 2: add citations and an instruction to refuse when context is missing.
• Step 3: create 30 test questions with known answers and measure retrieval recall.
• Step 4: add keyword search with BM25 and fuse results.
• Step 5: add a reranker and compare metrics.
• Step 6: wrap it in a FastAPI service with streaming.

Each step should improve a measured number, not just add code.

What is RAG analysis?

In AI, RAG analysis means evaluating a retrieval-augmented generation system: whether retrieval finds the right chunks, whether answers are faithful to them, and where failures come from. In project management, RAG analysis is a Red, Amber, Green status review.

For AI systems, analyse each stage: retrieval recall at k, precision of retrieved chunks, answer faithfulness, answer relevance and citation accuracy. Tools such as RAGAS automate some metrics, but read failures by hand too. Typical findings: bad chunking splits answers, embeddings miss exact codes, or the prompt allows guessing. In project reporting, a RAG analysis colours each workstream's status and explains the reasons and recovery plans.

How do I build RAG in Python?

Embed your chunks, embed the question, find the most similar chunks with cosine similarity, and put them in a prompt for a language model. A minimal version needs only a few libraries.

Also asked as: tutorial rag python

from sentence_transformers import SentenceTransformer
import numpy as np
model = SentenceTransformer("all-MiniLM-L6-v2")
chunks = open("handbook.txt", encoding="utf-8").read().split("\n\n")
emb = model.encode(chunks, normalize_embeddings=True)
q = model.encode(["What is the attendance rule?"], normalize_embeddings=True)[0]
top = [chunks[i] for i in np.argsort(emb @ q)[::-1][:4]]
prompt = "Answer only from these passages and cite them:\n" + "\n---\n".join(top)

Send prompt plus the question to any chat model. Then replace the list with a vector store and add evaluation.

How do I build RAG with LlamaIndex?

LlamaIndex can load documents from a folder, build a vector index and create a query engine in a few lines, then you customise the language model, embedding model, chunking and retriever.

Also asked as: llamaindex tutorial rag

from llama_index.core import VectorStoreIndex, SimpleDirectoryReader
docs = SimpleDirectoryReader("data").load_data()
index = VectorStoreIndex.from_documents(docs)
engine = index.as_query_engine(similarity_top_k=4)
print(engine.query("What are the hostel rules?"))

By default it uses an OpenAI model and needs an API key; configure other providers or local models through its settings. Next steps: persist the index to disk or a vector database, set a node parser for chunk sizes, return source nodes for citations, and evaluate with LlamaIndex's evaluation modules.

What are some RAG examples?

Common RAG examples include customer support bots answering from help articles, internal knowledge assistants over company documents, study assistants over lecture notes, legal and policy search with clause citations, and medical guideline lookup for clinicians.

Examples from my work: RAG.NextUpgrad answers with cited, streamed responses and a confidence gate that refuses weak-evidence answers; the OCR and Speech Workspace runs RAG scoped to one uploaded document with page citations; and DocuLens AI combines retrieval with schema validation for documents. The pattern is always the same: retrieve trustworthy context, constrain the model to it, and show sources so users can verify.

Where can I find RAG examples on GitHub?

Good starting repositories include LangChain's "RAG From Scratch" notebooks, the examples folders in LlamaIndex and LangChain, Hugging Face's open-source AI cookbook, and community collections such as "RAG Techniques", which demonstrates many advanced methods with notebooks.

Also asked as: rag examples github

When reading examples, check the date and library versions, because RAG tooling changes fast and older code may use deprecated interfaces. Prefer examples that include evaluation, not just a demo query. Run them on your own documents to see real behaviour. Then build your own repository with a clear README, architecture diagram and evaluation results; that is far more valuable for your portfolio than a forked example.

Is there a free RAG course?

Yes. DeepLearning.AI offers free short courses on RAG and advanced retrieval, LangChain Academy has free courses on LangChain and LangGraph, Hugging Face provides free cookbook tutorials, and many university lectures on information retrieval are online.

Also asked as: rag course free

A free learning plan:

• Take one short RAG course to learn the vocabulary.
• Follow the LangChain or LlamaIndex official RAG tutorial.
• Work through the "RAG From Scratch" notebooks.
• Build a project on your own documents with evaluation.
• Study hybrid search and reranking.

Courses give structure, but the project teaches the real lessons, such as messy PDFs and questions your retriever cannot answer.

Is there a free RAG course with a certificate?

Some platforms offer free completion badges or certificates, such as Microsoft Learn and Google Cloud Skills Boost for their AI learning paths, while many Coursera courses can be audited free with certificates available through payment or financial aid.

Also asked as: free rag course with certificate

Certificates help a little on a resume, but employers hiring for GenAI roles care much more about a working RAG project with evaluation results, a clear write-up and code they can read. If you take a certificate course, pair it with a deployed project. Check each platform's current terms, since free certificate offers change.

Is there a RAG pipeline course?

Yes. Several courses focus on full RAG pipelines, from ingestion to evaluation and deployment, including DeepLearning.AI's courses on building and evaluating advanced RAG, framework academies, and cloud provider training on managed RAG services.

Look for courses covering: document parsing and OCR, chunking strategies, embeddings, hybrid search, reranking, prompt design with citations, evaluation metrics, monitoring and cost control. Skip courses that only show a single demo query. After the course, rebuild the pipeline yourself on a new dataset; if you can explain each design decision and show metrics, you have learned it.

How do I make a rag quilt?

A rag quilt is a cosy quilt with exposed seams that fray into a soft, ragged texture. Cut squares of flannel or cotton in layers, sew them with seams facing outward, snip the seam allowances, and wash so the edges fluff up.

Also asked as: tutorial rag quilt

Steps:

• Cut matching squares for top, middle and back layers.
• Stack each set and sew an X through it to hold layers.
• Sew squares together with seam allowances on the front side.
• Sew around the outer edge, leaving an allowance.
• Snip every allowance at short intervals without cutting stitches.
• Wash and dry to fray the edges.

This rag has nothing to do with AI, but it shares the search term.

Pranjul Rathour

GenAI Engineer · India

GenAI engineer and AI product builder with 2+ years shipping production-grade AI systems: RAG pipelines, fine-tuned LLMs, hybrid retrieval and multi-modal apps across vision, speech and OCR, architected end to end from ingestion to deployment. Leads engineering for SCULT INDIA's 14-member team, founded the 500+ member TechVerse Enclave community and has mentored 200+ students. Three hackathon first prizes: Changethon 2025 (IIT Roorkee), Product Genesis at Vividhotsava 2025 (CSJMU) and BYTEBATTLE (MeetKats).

Open to GenAI roles, hackathon judging, mentorship sessions and guest talks at colleges. For campus sessions the inviting institution covers the visit end to end: travel (a First AC train ticket for journeys up to five hours, a flight for anything longer), accommodation, meals, and pickup and drop; the honorarium and workshop fees are negotiable on WhatsApp or on the call, and requests are welcome from institutions that can cover the visit this way. Email pranjulrathour41@gmail.com.

PortfolioInvite me to your campusHire meLinkedInGitHubXInstagramThreadsFacebookBlueskyDev.toBloggerHashnodeSCULTEmail