关于我Warp — the steady threads
Hello,I’m Yanqing Li.
14 年测试与质量工程经验，当前聚焦 AI 产品质量、自动化与持续监测、质量平台和智能体工程实践。
The profile text below is written in Chinese.
- Based in
- Beijing, China
- Experience
- 14 years in testing and quality engineering, since 2012
我是李彦卿，在北京从事测试开发与质量工程工作，现任汽车之家高级测试开发工程师。14 年来覆盖 Web、App、H5、小程序、插件与后台系统，持续建设自动化框架、接口回归和质量工具。
自研 PulseGuard 已在公司多个项目中用于前端页面与 API 探活；个人持续研发 CaseFlux，并开源 AI Playwright Framework、Rootloom 和衡木，实践测试资产、真实执行、失败诊断与工程决策。
本站通过项目、案例、文章和简历呈现这些实践，并分别注明公司应用、个人研发与开源状态。
Warp — 2012 to now
Career
2012 to now, 3 roles, all of them in testing and quality engineering. The warp goes on the loom before the weft has anywhere to pass.
Full résumé- Earlier careerTest Engineer | Renren, Lashou, Daojia, Shouqi2012.04 — 2017.10
- MockingBot (墨刀)Test Engineer / Test Development Engineer2017.11 — 2021.01
- Lanhu (蓝湖)Test Development Engineer / Test Lead2021.02 — 2024.08
Method — how I work
How I work
From delivery work to engineering review, these are the capabilities I come back to daily. The project detail lives in Work and Cases, if you want to read them side by side.
-
Business quality and delivery
Requirements review, test strategy, release gating and defect retrospectives, covering the exception paths and quality risks in AI marketing, auto-reply and multi-platform products.
-
Automation and continuous monitoring
Built Playwright / pytest automation, API regression and CI gates, and put my own PulseGuard to work on front-end page and API health checks across several company projects.
-
Quality platform engineering
Built the whole thing myself — requirements modelling, a React front end, a Python / FastAPI back end, execution nodes and Docker deployment — turning scattered scripts into reusable, traceable test assets.
-
AI testing and agent practice
Working in CaseFlux and the AI Playwright Framework on natural-language cases, real browser execution, failure diagnosis and repair review, with attention to generation quality, assertions and execution evidence.
-
Engineering review and technical decisions
Using Rootloom and Hengmu to practise risk grading, root-cause verification, architecture review, option trade-offs and deterministic quality gates, so engineering conclusions rest on real verification.
Weft — what I’m exploring
What I’m working on
Quality management is the day job; these are the weft threads that keep crossing it — hands-on experiments in personal and open-source projects, and the process written up as articles.
- 人工智能产品的评测、回归与线上质量
- 编码智能体的风险、根因、范围与验证证据
- 架构发现、技术决策与确定性质量门禁
- Playwright / pytest 自动化与质量平台
Next — a direction, not a destination
Next, I want to go deeper into AI testing and development.
That is a direction, not a place I have reached. What I can show now is the projects in progress and the articles already written; new work will keep being recorded here.
Knot — where to next
Continue from here
结 is a knot, and also to make someone’s acquaintance. To see what I have built, read what I have written, or just talk — all of it starts here.
- Work 10 projects: in-house tools, products in development, and open-source projects, each stating its real status.
- Writing 6 articles: {prefix}on AI testing, agent engineering, and the judgements and trade-offs in between. These are in Chinese.in Chinese
- Cases 4, each setting out the problem, my own contribution, the key decisions and the evidence. These are in Chinese.in Chinese
- Résumé The full work history, selected projects and skills, with a PDF to download.