跪拜 Guibai
← Back to the summary

A Free, Open-Source Roadmap for Becoming an AI-Era Test Developer

Hello everyone, I'm Kuangshi.

In the past two years, AI has flipped software testing on its head. Test cases can be drafted by AI, scripts can be written with AI's help, and even failure logs can be triaged by AI. Suddenly, manual testers are panicking, functional testers are panicking, and newcomers wanting to enter the field are panicking too. Everyone is asking the same question: if AI can do all this, is testing still a viable career?

Yes, it is, and there are more opportunities than before. Once AI takes over the repetitive work of writing cases, scripting, and sifting through logs, people are freed up to do the critical work: deciding whether a release has been tested enough, whether quality is slipping, and pinpointing which layer a problem sits in. These decisions still require human judgment.

So, does the testing field lack talent? Yes, it lacks people who understand both testing and AI. To become such a person, you need to be strong in two areas: the traditional fundamentals of test development, and the new skills of the AI era. I searched for a learning path that clearly strings these two together and found none. So I laid one out myself — from zero foundation to test development engineer, covering how to build the fundamentals and how to train the new skills, all in one path.

This is the "AI Test Development Learning Path Compendium", officially released as v1 on August 28th, completely free and open-source. It includes 13 chapters, 10 stages, 4 progressive hands-on projects, a set of interview guides, and a companion online website that updates synchronously.

Project open-source address: https://github.com/zhoujinjian/ai-testing-guide

 https://github.com/zhoujinjian/ai-testing-guide

If you find it useful, please Star it, and feel free to share it with friends around you who are switching careers or feeling anxious about testing.

My only goal is this: even if you have zero foundation, you can follow this path to first build a panoramic understanding of AI test development, then train both the traditional SDET fundamentals and the new AI tracks and skills, and strive to land a satisfying SDET offer as soon as possible.

To put it bluntly: a free, systematic, and background-customized SDET learning path that hits all three criteria — I haven't seen a second one on the market.

Besides the open-source repo, this path also has an online reading version. Visit the "Learning Path" section of the "AI Test Development Navigator" website, find "2026 Latest Test Development Learning Path", click to open, and you can read it online (for free).

Online reading address: https://www.testfather.cn/learning-paths

What is Test Development (SDET)?

Simply put, test development is "doing quality through engineering." Writing test cases and clicking through pages is just the entry level. A real SDET must be able to write test code, build automation frameworks, integrate testing into CI pipelines, do performance stress testing and build quality platforms, and when problems arise, use logs and data to pinpoint which layer is stuck.

The AI era has added two new dimensions to this role. One is using AI to boost testing efficiency: generating test cases, writing scripts, reviewing code, and triaging failures. The other is evaluating AI applications themselves: how to test, verify, and set quality gates for new products like knowledge bases, RAG, and Agents.

The former is the SDET's fundamental skill; the latter is the new requirement of the AI era for SDETs.

Entering SDET: Three Hurdles to Overcome First

The first is the skill hurdle — you need to know more than you'd think. Open any SDET job description, and the requirements list is long: a programming language, interface and UI automation, performance stress testing, Linux, MySQL, CI/CD, and some even explicitly ask for the ability to build a quality platform. For manual testers, fresh graduates, and career-switchers, this is a whole wall. Daytime work or classes already drain most of your energy; at night, when you try to catch up, you find dozens of learning paths online, each saying something different. Today you learn Postman, tomorrow JMeter, and the more you patch, the more chaotic it gets. The wall is still that wall.

The second is the practice hurdle — without being in the industry, there's nowhere to practice. Job postings all require project experience, but real projects and environments are locked inside companies. Since you haven't entered the industry yet, you naturally can't access them. Self-study can only produce toy projects: data is self-fabricated, traffic is self-simulated. When an interviewer probes two questions deep into real scenarios, you're easily exposed.

The third is the interview hurdle — both written tests and interviews dig deeper. Written tests include coding and algorithm questions; SDET is, after all, a technical role. Interviewers are not satisfied with "knowing how to use" something. Why is the framework layered this way? Why do scripts break every few days? How do you locate the cause when they break? How is automation integrated into CI? Dig two layers down, and those who only followed tutorials can't keep up.

These three hurdles look different, but they test the same thing. Tools are quick to pick up — anyone can learn them. Tools alone won't get you past these three hurdles. What's really being tested are three abilities: writing maintainable test code, running automation in CI and daily regression, and using logs and data to explain a failure. The foundation of this learning path is to break these three things down and practice them in every stage.

And the bar is still rising. People who can use AI finish in one day what used to take three; those who can't are still hand-writing assertions field by field. AI-driven efficiency is becoming an implicit requirement in hiring. On the other side, the AI applications companies are building urgently need people to test them. How do you assert when answers aren't fixed? How do you set safety red lines? Most teams have no ready-made answers. The old hurdles haven't lowered, and new ones have been erected. Whoever masters them first will be the first to catch this wave of new demand.

What's Inside This Learning Path?

This path covers 13 stages of SDET advancement, a complete growth map from functional testing to quality engineering: first, build solid fundamentals, then leverage AI for efficiency.

The entire path is arranged by cultivation realms: Qi Refining, Foundation Building, Core Formation, Nascent Soul, Spirit Severing, all the way to Tribulation and Ascension. At the end of each chapter, there's a "Self-Check Checklist." Only when you can independently answer all the questions inside is that chapter considered passed. The original text in the path says: "Realms are not advanced by meditation, but by surviving the tribulation of each chapter's self-check checklist. Skipping stages leads to deviation; bookmarking without learning destroys the foundation of your Dao."

Who Is This For?

1) Students and fresh graduates. Zero foundation or career-switchers are fine. Follow from Chapter 1 to Chapter 13 in order, and by graduation, you'll have one or two projects you can talk about — your resume won't be hollow.

2) Functional/business test engineers. Your advantage is business understanding and testing mindset. The path highlights what you need to supplement: programming, engineering fundamentals, and interface automation. The two AI chapters are your best card for transitioning to SDET.

3) Programmers and backend developers switching to SDET. You can basically skip the coding part. Focus on supplementing test case design and assertion thinking. In interviews, switch your narrative from "what feature I developed" to "what quality I ensured."

4) In-service testers wanting to advance. You already know some automation but are stuck at a bottleneck. Based on your target role, selectively study the performance, quality platform, and precision testing chapters to aim for higher positions and salaries.

5) People wanting to enter the AI testing direction. Whether you have a testing or development background, Chapters 10 and 11 are written for you: use AI to boost testing efficiency, or directly pivot to doing AI application evaluation.

Where to Start Learning?

Different backgrounds don't need to read from Chapter 1 to Chapter 13. The path overview provides three customized tracks.

Functional testing to SDET: Focus on supplementing programming, engineering fundamentals, and interface automation. Start with Project One (Interface Automation Framework); it connects most naturally with your business experience.

Backend to SDET: You can basically skip the first three stages. Focus on test case design and assertion design, then head straight for quality platforms and AI application evaluation.

Campus recruitment, zero foundation: Follow the order, don't skip, and don't avoid algorithms. SDET is a technical role; encountering algorithm questions in written tests and first-round interviews is normal. Start Project One in your sophomore year or first year of grad school, and by graduation, you'll have two or three complete projects in hand.

Besides these three tracks, there are two common suggestions. Read the two AI chapters early — building judgment doesn't require knowing how to code first. Starting from interface automation, you can directly use AI to boost efficiency, learning and using it simultaneously. The path includes a 3-to-6-month pacing table, calculated based on 2 to 4 hours of daily investment, listing what deliverables should be produced each month. If time is tight, prioritize programming language and interface automation.

Hands-on practice is the best teacher. Don't wait until you've "finished learning everything" to start. Begin working on projects after completing the first five stages, filling gaps as you go.

Final Words

This learning path is open-sourced under the CC BY-NC-SA license. Personal learning and blog reposts with attribution are freely allowed, just note the source. Using it to run paid columns or paid training is not permitted by the license; if seen, rights will be defended.

One person's energy is limited; the content inevitably has shortcomings. The path will be continuously updated, and Issues and PRs are welcome for co-construction.

Project open-source address: https://github.com/zhoujinjian/ai-testing-guide

 https://github.com/zhoujinjian/ai-testing-guide

If this learning path helps you, go to GitHub and give it a Star ⭐️. Your recognition is the motivation for continuous updates.

Don't hesitate. Open Chapter 1 now; it takes just an hour or two to read. After reading, you'll know whether this path suits you. Start from the Qi Refining stage, step by step, and cultivate all the way to Tribulation and Ascension 🚀

Some Supporting Resources

👉🏻 Online learning path with progress tracking https://www.testfather.cn/learning-paths

👉🏻 Staged practical tutorials, matched to the path chapters https://www.testfather.cn/tutorials

👉🏻 High-frequency SDET interview questions for online practice https://www.testfather.cn/interviews

👉🏻 Skill self-assessment to locate weak areas https://www.testfather.cn/interviews/assessment

👉🏻 Test development job resume templates https://www.testfather.cn/resume-templates

👉🏻 AI testing efficiency Skills and prompt tools https://www.testfather.cn/skills