Who this course is for
Manual testers who want to stay employable as their product grows an AI feature, automation testers and SDETs who need to write assertions against a model's output, and QA leads who have been handed an LLM feature and asked for a test plan. You need working knowledge of software testing. You do not need a machine-learning background, and Python is taught from zero in the companion Python and DSA track, which the same subscription covers.
What the 16 modules cover
Foundations and evaluation
What a large language model actually does, why the same prompt gives different answers, and how to score an answer you cannot diff. Prompt engineering from a tester's point of view, and building a ground-truth set you can measure against.
Red-team and safety testing
Hallucination detection, prompt injection and jailbreaks, data leakage, bias and toxicity. Writing adversarial cases on purpose, and deciding what counts as a defect rather than a model being a model.
RAG systems
Retrieval quality, chunking and embedding choices, citation faithfulness, and knowledge-base poisoning — where a single planted document changes what the system tells every user.
Agents, multi-agent systems and MCP
Testing a system that decides its own next step: tool selection, refusals, budget and loop limits, hand-offs between agents, and Model Context Protocol tool calls.
Automation
Promptfoo, DeepEval and RAGAS wired into a suite that runs in CI, plus observability and tracing so a failure in production is diagnosable rather than just reported.
Two capstone projects
Not a walkthrough video — a running application you attack. See the projects section below.
All 16 modules, in order
The whole syllabus is here to read before you decide anything — no sign-up, no email, no "request the curriculum". Module 1 is free to work through in full; the rest need a subscription. About 93 hours of material in total.
You test real running applications, not slides
Three applications are hosted for you, and you work them from a browser with no installation. This is the part that separates the course from a video playlist: you are looking at a real system's actual output when you decide whether it is a defect.
Project 1 — the RAG application
A travel-planning assistant in five versions, each one broken in a different, deliberate way. You run the retrieval, inspect the vector store, plant a poisoned document and watch it change the answers, then run red- and blue-team suites against it and read the trace of everything you did.
Project 2 — the multi-agent application
A booking system built from several agents with real tool calls over MCP. You test hand-offs, watch an agent refuse a call it is not allowed to make, break the budget check, and tell the difference between a control working and a component failing.
Certification
Three separate exams — Basic, Advanced and Expert. Each one you pass earns its own certificate carrying a unique number, and anyone can check that number against this site without an account, which is what makes it worth putting on a CV. Nothing is awarded for finishing the videos; the exams are the bar.
What it costs
Module 1 is free once you have an account — no card, nothing else to set up. After that one subscription unlocks the rest of this course and the whole Python and DSA track: ₹499 for a month, ₹1,199 for three months, ₹3,999 for a year. Nothing renews on its own, and buying again while you still have time left adds days rather than replacing them. Full pricing.
Common questions
How this compares with ISTQB CT-AI, whether you need Python first, and what an AI testing career actually looks like are all answered on the FAQ page.