That question breaks every test you know how to write. This course teaches you to answer it the way AI teams do: a pass rate, a range, a threshold, and the cases that failed. Twenty-three weeks, from your first terminal command to a release assessment you present on video.
In plain words: the bot answered the in-scope questions right 17 times out of 20. The bar was 18. It is not ready to ship, and this output says exactly why.
What the course is, how it works, and exactly what arrives in your inbox when you buy. Under ten minutes together. Until they are up, each tile lists what its video covers, and the syllabus says the rest.
Every team is putting an AI assistant in front of customers. None of them can test it the old way, because the same question gets a different answer each time. They need a tester who can measure that. Not a data scientist.
One input, one expected result, one answer.
Test design, exploration, bug reports, judgment about severity.
Enough Python to run the tools, enough statistics to trust a number, and the new failure types.
A portfolio that proves it, and a title with "AI" in it.
You keep hearing these words and nobody stops to explain them. The course starts by answering 22 of them in plain language, in your first 15 minutes. Guess first, then open one.
A model is a program that learned its behaviour from examples instead of being written as rules. Nobody typed "if the guest asks about dogs, say 25". It read a huge amount of text and learned what a sensible reply looks like. That is why you cannot find the line of code that caused a wrong answer.
The unit a model reads and writes: a word or a piece of a word. "Cancellation" might be two or three tokens. It matters to a tester because cost, speed, and size limits are all counted in tokens, not in words.
Retrieval-augmented generation. A model knows nothing about your company's documents and its general knowledge stops at a cutoff date. RAG fixes both: when a question arrives, the system first searches your documents for the relevant passages, then hands them to the model and says "answer from these". Without it the bot guesses. With it, the bot can still go wrong in two new places: the search can fetch the wrong passage, or the model can ignore the right one. You will test both.
The model does not look up an answer. It builds one, a token at a time, and at each step it picks from several likely options with a little randomness. A setting called temperature controls how much. So one run tells you very little, and that single fact is why AI testing is a different job.
No. You need enough Python to loop over a list of questions, call an API, and write a line that says "this must be true". That is a few weeks of learning, and Stage A teaches exactly that and stops.
AI-testing titles are still a small slice of QA jobs, but the fastest-growing and best-paid slice. Most of the demand hides inside SDET and QA roles that now ask for LLM, RAG, or evaluation skills. Those are Stages A to C.
| Role | India, per year | United States | What gets you there |
|---|---|---|---|
| Manual / functional QA | ₹3–7 L₹14–20 L at lead level | $55–110k | Your starting point. Test design, exploration, and defect reporting all carry over. |
| SDET / automation engineer | ₹10–28 L₹30–55 L at product companies | $95–195k | Stage A: Python, pytest, Playwright, CI. |
| AI Test Engineer / AI Quality Engineer | 20–35% above SDET₹42–65 L reported for specialists | $180–240k | Stages A to C: eval suites, golden sets, rubrics, judges, a gate in CI. |
| LLM Evaluation Engineer | ₹15–40 L+ | $160k average | Modules 7 and 8: promptfoo, DeepEval, RAGAS, Langfuse, retrieval and agent testing. |
| AI Red Team Analyst | ₹22–50 L₹50–80 L at lead level | $80–220k | Module 9: prompt injection, leakage, guardrails, OWASP LLM Top 10, PyRIT and Garak. |
| Prompt QA / prompt engineer | ₹4.6–6 L average₹25–60 L when paired with code | $100k+ | Module 6, plus Stage A. Prompting without code is the one path that plateaus. |
Compiled 28 September 2026 from LinkedIn, Indeed, and Glassdoor listings and three 2026 QA market reports (SoftwareTestPilot, InterviewStack, ScrollTest) and the KnowledgeHut red-team salary guide. These are ranges, not promises: product companies in Bengaluru and Hyderabad pay at the top of each band, services firms at the bottom, and the AI-specialist figures rest on a few hundred salary reports. No course can guarantee a salary.
Every module ends with something in your GitHub repository that a hiring manager can open, run, and read. These are the ones interviews are built around.
Ten Playwright tests against a live demo shop, twenty green runs in a row, three real defects documented.
One chatbot question, twenty runs, a table of what varied, and thresholds a product owner could sign.
A 40-question golden set, a judge you calibrated by hand, and a gate that blocks a bad change from merging.
Retrieval measured on its own, answers checked claim by claim, and a test that catches a destructive action before it happens.
Thirty attack cases across the OWASP LLM Top 10, an automated sweep, and findings with severities and retests.
The capstone: full suite in CI, red-team, a two-page go or no-go, and a ten-minute recorded walkthrough.
Harbour View Hotel has a support bot that answers policy questions and a concierge agent that can cancel bookings. You build both from parts you write yourself, so you understand every line.
You get the code. You do not get the list. Your evaluation suite has to find them, and the reveal at the end shows your pass rates moving as each one is switched off.
Stage A has no AI in it on purpose: seven weeks of foundations so every tool later is something you can run and read. Already write Python? Skip Modules 2 and 3 and finish in nineteen weeks.
Every module has a full page of lesson content: objectives, teaching notes, student material, checkpoints with answer keys, and labs with a definition of done. Read the complete syllabus.
Every lesson has the same five parts, so you always know what "done" looks like.
Two to four things you can be tested on.
Runnable in every lesson, checked before it ships.
Three questions and an answer key.
A definition of done and points.
Every AI call prints its price. Under $10 for the whole course.
Chandru builds and tests software products, and wrote every lesson, lab, and line of practice code in Zero to AI Tester. It is the path Chandru wanted and could not find: one that starts at the terminal and ends at a release assessment.
Built for testers who are told AI will replace them, by someone who thinks the opposite: the people who already know how to design a test are the people AI teams need most.
Pay once by UPI or card. The whole course arrives by email within a minute, as pages you open in your browser. No subscription, no expiry.
All sales are final. It is a digital product delivered in full the moment you pay, so there are no refunds. Unsure? Email a question first.
As a pack of web pages, by email, within a minute of paying. Unzip it, open the start page in any browser, and everything is there offline, forever. When a lesson is updated you get a new pack at no charge.
No. All sales are final. It is a digital product delivered in full the moment you pay, which is why the syllabus, the videos, and the first 15 minutes of the course are free before you pay. Email a question first if anything is unclear.
About ten dollars of AI credit, bought from the model provider. Every script prints what it spent. Every other tool in the course is free and open source.
About seven hours a week for 23 weeks. Two weekday evenings for lessons and one weekend block for the lab. Already write Python? Skip Modules 2 and 3 and finish in about 19.
Yes. That is who it is written for. Module 2 teaches Python from the first line with every example about testing. You will be writing test code by week four and driving a browser by week six.
Either. Every command is shown for both. You need a laptop with 8 GB of RAM or more; Module 0 installs VS Code, Python, and Git and checks them for you.
Python, pytest, Playwright, GitHub Actions, the Claude API, promptfoo, DeepEval, RAGAS, Langfuse, and Claude Code with the Playwright MCP server. Garak, PyRIT, and Great Expectations as a survey. The principles come first, because the tools will change.
No certificate, on purpose. You leave with a public portfolio of seven projects and a recorded release assessment, which is what interviews are built around. Module 10 maps the course onto the ISTQB CT-AI syllabus if an employer filters on it.
No. Only the practice app on your own machine and a sandbox API built for testing. The module opens with the authorisation rule and the terms require it.
Every checkpoint has an answer key and every lab a rubric. Module 0 teaches the fifteen-minute rule and a four-line "what I ran, what happened" template. Beyond that, email works.
More courses are planned in the same style: from zero, real code, a broken practice product, a portfolio at the end. Tick what you would buy and you hear first, with an early price.
No newsletter, no weekly emails. One message when a course you ticked launches, and nothing else.
That sentence is the job. This is the course that gets you to it.