We earn a commission if you buy through links on this page. How we're funded.
Quick answer
Retrieval Augmented Generation (RAG) by DeepLearning.AI is a hands-on, intermediate Coursera course for developers who want to build RAG systems properly, and we rate it 4.2 out of 5. Its five modules cover keyword and semantic search, vector databases, chunking and reranking, text generation with language models, and running RAG in production, with a graded programming assignment in each, at 31 hours by Coursera's module cards. It needs intermediate Python, and it ends in a single-course certificate rather than a professional one. For a longer programme that ends in a professional certificate, IBM's RAG and Agentic AI Professional Certificate is the alternative we would weigh.
Why we score it 4.2 / 5
An intermediate DeepLearning.AI course for learners with intermediate Python: keyword and semantic search, vector databases, chunking and reranking, text generation with language models, and running RAG systems in production, across five modules and 31 hours by Coursera's module cards. We value the build, five graded 180-minute programming assignments that implement retrievers, a vector-database RAG system and a chatbot, on current material such as agentic and multimodal RAG. What holds the score down is the price and the credential: it is bought per course or through Coursera Plus at a price Coursera sets by country, for a single-course certificate rather than a professional one.
4.7 / 5 learning score — our reading of the published curriculum4.0 / 5 credential score — our reading of who issues it
Curriculum currency 4.6 · Completion realism 4.6 · Skill value 4.8 · Employer recognition 4.0 · Cost & value 3.0 · Salary impact 4.0 — the score is the average of these six, each out of five.
Scored with AI assistance against our published rubric, from what the provider publishes; the editor is responsible for the rubric and for every published score. What our scores can and cannot tell you.
Provider facts for this entry were last checked on 2026-10-09.
Check price & enrol on Coursera →
This review covers what DeepLearning.AI's Retrieval Augmented Generation (RAG) course teaches across its five modules, what it asks of you and costs, how it compares with the other RAG and LLM-engineering courses we review, and who should take something else. Every fact in it comes from the course's Coursera page as it stood on 9 October 2026.
What is DeepLearning.AI's RAG course?
Coursera titles it simply “Retrieval Augmented Generation (RAG)”. It is a single course from DeepLearning.AI, not a programme: five modules, no member courses, ending in a shareable Coursera course certificate. Its instructor is Zain Hasan, and its first module opens with a conversation with Andrew Ng.
Retrieval-augmented generation is the usual technique behind question-answering systems built on an organisation's own documents: before a language model answers, the system retrieves the passages most relevant to the question and gives them to the model, so the answer rests on your data rather than on what the model happened to learn. A demo takes an afternoon. A system that retrieves the right passages for real questions, at a cost and speed you can live with, is the hard part, and this course is built around that part.
Coursera lists it as Intermediate, and its recommended experience is “Intermediate Python skills required; basic knowledge of generative AI and high school–level math is helpful.” On 9 October 2026 the page showed 56,384 learners already enrolled and an average of 4.8 stars from 220 reviews. The page's own data gives a launch date of 14 July 2025, and the page shows no update date.
What you learn, module by module
Coursera lists the five modules in this order, under these titles:
- RAG Overview
- Information Retrieval and Search Foundations
- Information Retrieval with Vector Databases
- LLMs and Text Generation
- RAG Systems in Production
Across those modules, Coursera's outlines name a wide set of techniques. On retrieval: keyword search with TF-IDF and BM25, semantic search and embedding models, hybrid search, metadata filtering and evaluating retrieval. On vector databases: approximate nearest-neighbour search, chunking, query parsing, cross-encoders and ColBERT, and reranking. On generation: transformer architecture, how language models sample their output, prompt engineering, hallucinations, agentic RAG, and when to use RAG rather than fine-tuning. On production: evaluation, logging and observability, quantisation, the trade-offs between cost and latency, security, and multimodal RAG. Coursera's own summary puts the aim plainly: “How to weigh tradeoffs between cost, speed, and quality to choose the right techniques for each component of a RAG system”.
The practice is the reason to take it. Each module carries a graded programming assignment, listed at 180 minutes each: an introduction to RAG systems, implementing retriever functions, building a RAG system with a vector database, developing a RAG-based chatbot and improving that chatbot. Nine ungraded labs and a short quiz per module sit alongside them. The labs use the Weaviate vector database's API, and the course description names open-source language models hosted by Together AI.
The emphasis on retrieval is the right one. Most RAG systems that disappoint do so because they fetch the wrong passages, not because the model writes badly, and this course spends two of its five modules on retrieval, and its outlines name the evaluation of retrieval as well as of the whole system. One practical note: a reading in the course is titled “[IMPORTANT] Reminder about End of Access to Lab Notebooks”, so keep your own copies of the lab work.
Cost, time and what you need first
Time. DeepLearning.AI's Retrieval Augmented Generation (RAG) course runs to 31 hours by Coursera's module cards. Coursera's headline gives a pace instead, three weeks at 10 hours a week. The five graded assignments account for a large share of those hours, so a learner who skips the ungraded labs will finish sooner and learn less.
Cost. It is bought per course or through a Coursera Plus subscription, and Coursera prices it by country; the page states no price, and the button reads “Enroll now”. If you already hold Coursera Plus, it is included. Coursera's financial aid is a discount you apply for from the course page; Coursera says a decision can take up to 16 days, and starting a free trial cancels a pending application.
What you need first. Take the recommended experience literally. If you can write functions and classes, work with lists and dictionaries, install a package, read a traceback and have called a language model's API, you are ready. If not, the graded assignments will mean learning Python and RAG at once; start with Python and a first LLM application.
Certificate. A shareable Coursera course certificate from DeepLearning.AI, added to your Accomplishments page when you finish. It records completion; there is no proctored exam and no vendor certification attached.
Check the price & enrol on Coursera →How we checked this. We list the cost as Paid: bought per course or through a Coursera Plus subscription, priced by country. Source: Coursera course page coursera.org/learn/retrieval-augmented-generation-rag, read server-side 2026-10-09: the button reads "Enroll now" and the page shows "Included with Coursera Plus" and "Financial aid available"; it states no price. Coursera prices by country, and this course's enrol screen has not been read in a browser, so no figure is published. Where we give a figure, the source says when we read it on the provider's page. We re-read prices by hand and publish no figure we cannot source — where a provider prices regionally, we say so rather than quote a number that is wrong for most readers.
Pros and cons
✓ What we liked
- You build the components yourself: a graded programming assignment in every module
- Retrieval taught properly: BM25, semantic and hybrid search, reranking and retrieval evaluation
- Goes past the demo: evaluation, observability, cost and latency, and security
- Current material, including agentic and multimodal RAG, and included in Coursera Plus
✕ What to keep in mind
- Needs intermediate Python before you start
- A single-course certificate, not a professional certificate
- Labs use the Weaviate API, which may not be the vector database you use
- The page states no price; Coursera prices it by country
Not sure this is the right one for you?
Tell the picker about your background and what you want the certificate to do, and it narrows the list to the one or two courses we would start with. It suggests only our affiliate partners’ courses, and says so before it suggests anything.
Try the AI Certification Picker →Who should take it, and who should not
Take it if you are a developer or data scientist who has called a language model's API and now needs to answer questions over your own documents; if you have built a RAG demo with a framework and want to understand why it retrieves the wrong things; or if you are moving into AI engineering and want retrieval, the least glamorous part of the stack, done properly.
Skip it if you do not write Python yet: start there. Skip it if what you need is a credential an employer filters on, where IBM's RAG and Agentic AI Professional Certificate ends in a professional certificate. And if you want to understand RAG without building it, our explainer What Is RAG? covers the idea in plain English.
How it compares with the other RAG and LLM courses we score
The table below compares five RAG and LLM-engineering courses on provider, level, time, how you pay and our rating. All five are Intermediate and use Python; they run from about 3 hours to 101.
| Course | Provider | Level | Time | How you pay | Our rating | Enrol |
|---|---|---|---|---|---|---|
| DeepLearning.AI Retrieval Augmented Generation (RAG) | DeepLearning.AI (Coursera) | Intermediate | ~31 hrs | Per course or Coursera Plus | 4.2 / 5 | Coursera → |
| IBM RAG and Agentic AI Professional Certificate | IBM (Coursera) | Intermediate | ~101 hrs | Coursera subscription or Plus | 3.9 / 5 | Coursera → |
| Agentic AI (DeepLearning.AI) | DeepLearning.AI (Coursera) | Intermediate | ~23 hrs | Per course; not in Coursera Plus | 4.1 / 5 | Coursera → |
| AI Engineer Core Track: LLM Engineering, RAG, QLoRA, Agents | Udemy | Intermediate | ~33 hrs | Single purchase | 4.1 / 5 | Udemy → |
| Retrieval Augmented Generation (RAG) with LangChain | DataCamp | Intermediate | ~3 hrs | DataCamp subscription | 3.4 / 5 | DataCamp → |
IBM RAG and Agentic AI. A ten-course Professional Certificate that covers RAG with vector databases and advanced retrievers, then multimodal apps, AI agents with LangGraph and other frameworks, and the Model Context Protocol, ending in a capstone. It is about three times the length and ends in a professional certificate; this course goes deeper on retrieval itself in far fewer hours.
Agentic AI (DeepLearning.AI). The natural companion rather than a rival: Andrew Ng's course builds agent patterns in plain Python and has no module on retrieval, while this one covers retrieval and touches agentic RAG. Taken together they cover both retrieval and agents. Note that Coursera Plus includes this course but not that one.
Udemy's AI Engineer Core Track. A project-led course across the whole of LLM engineering, from building chatbots to fine-tuning and agents, with one part on RAG. Choose it for breadth and a one-off purchase; choose this course for depth on retrieval and for graded assignments.
DataCamp's Retrieval Augmented Generation (RAG) with LangChain. A three-hour course that builds RAG with LangChain and ends on evaluation with Ragas and on Graph RAG. It is the quick, framework-first introduction; this course is the thorough one.
Is DeepLearning.AI's RAG course worth it?
Yes, for the developer it is built for. We rate it 4.2 out of 5: it is current, it treats retrieval as the engineering problem it is, it carries the work through to production concerns such as evaluation, observability, cost and security, and its five graded assignments make you write the code rather than watch it written. What holds the score down is the price and the credential: Coursera sets the price by country and the page states none, and you finish with a single-course certificate rather than a professional one.
As with any engineering course, the certificate is a receipt. The thing to show an employer is a RAG system you built on documents you chose, with a short write-up of how you measured its retrieval and what you changed. If you want a professional certificate for the same subject, IBM's RAG and Agentic AI Professional Certificate is the alternative we would weigh. Our guide to the best RAG courses and certifications covers the wider field.
Ready to start?
Paid through Coursera rather than through the provider, by subscription or per course. Coursera prices by country: its pricing page shows the Coursera Plus plans and the price for your country.
Frequently asked questions
Is DeepLearning.AI's RAG course worth it?
Yes, if you already write intermediate Python and want to build retrieval-augmented generation systems that work on real questions. We rate it 4.2 out of 5. It teaches retrieval properly, from BM25 and semantic search to reranking and evaluation, takes you into production concerns such as observability, cost and security, and has a graded programming assignment in each of its five modules.
What holds it back is the credential and the price. It ends in a single-course certificate, which records completion rather than proving skill, and Coursera sets its price by country without stating it on the page.
So take it for the skill, and build something of your own alongside it: a RAG system over documents you chose, with a note on how you measured its retrieval, is what an interviewer will want to discuss.
How long does the DeepLearning.AI RAG course take?
It runs to 31 hours by Coursera's module cards, across five modules. Coursera's headline gives a pace rather than a length, three weeks at 10 hours a week.
The five graded programming assignments are listed at 180 minutes each, and there are nine ungraded labs besides, so much of the time goes into code rather than video. Skipping the ungraded labs gets you to the end sooner, but they are where the techniques become habits.
Plan by the hours rather than the weeks, and leave room for the graded assignments, which are the longest single items in the course.
What do I need to know before taking the RAG course?
Intermediate Python. Coursera's recommended experience reads: “Intermediate Python skills required; basic knowledge of generative AI and high school–level math is helpful.”
A fair test: if you can write functions and classes, work with lists and dictionaries, install a package, read a traceback and have sent a prompt to a language model through its API, you are ready.
If you are not there yet, learn Python and build one small LLM application first. If you want the idea before the code, our explainer What Is RAG? covers it without programming.
Is the DeepLearning.AI RAG course included in Coursera Plus?
Yes. Coursera's page shows “Included with Coursera Plus”, and its own data marks it as part of Plus (checked 9 October 2026). You can also buy it on its own; Coursera prices it by country, and the page states no figure.
That differs from DeepLearning.AI's Agentic AI course, which Coursera Plus does not include, so check the “Included with” line on each course page rather than assuming.
If the price is the barrier, Coursera's financial aid is a discount you apply for from the course page, and Coursera says a decision can take up to 16 days.
DeepLearning.AI's RAG course or IBM's RAG and Agentic AI certificate?
Choose DeepLearning.AI's course if you want to understand retrieval deeply in about 31 hours by Coursera's module cards, with a graded programming assignment in every module. Choose IBM's RAG and Agentic AI Professional Certificate if you want a broader programme that adds multimodal apps, AI agents and the Model Context Protocol, and ends in a professional certificate.
IBM's runs to about three times the hours, and both are included in Coursera Plus. We score DeepLearning.AI's course higher, mainly because it is far more finishable and its practice goes deeper.
If you have the time for both, take this course first: the retrieval it teaches makes the RAG courses in IBM's programme faster and easier to follow.
Keeping this current. Course formats, prices, and certification exam fees change and vary by region. We review our guides regularly, and we always recommend confirming the specifics on the provider's official page before you enrol.