Everything you need to go from manual QA to AI-powered test pipelines. Built and maintained by Vaishnavi Duvey — QA Manager & AI Testing

💡 Why I built this

Someone asked me: "Is there any way I could learn how to build these?" So I built a free learning kit. This repo started from a LinkedIn post that got 10,378 impressions — proof that QA engineers want practical AI content, not theory. Everything here is what I actually built and learned, shared openly so you don't have to figure it out alone.

"This is the kind of AI-in-QA content that feels practical instead of theoretical." — Comment from the LinkedIn community

🚀 How to use this repo If you're a beginner:

Start with the 5-week roadmap below — follow it week by week Read the architecture section — plain English explanation Run orchestrator.py — see AI generate real tests in 3 minutes

If you're intermediate:

Clone the repo and run orchestrator.py immediately Swap the story variable with your own user story Update standards with your team's conventions Star the repo — Stage 2 + 3 dropping this week

If you're advanced:

Fork the repo and contribute MCP connectors or RAG improvements Open a PR — reviewed within 24 hours Tag me on LinkedIn when you ship something with it!

This repo is a free, complete kit for QA engineers who want to build AI-powered testing pipelines using:

• 🤖 Claude AI (Anthropic) — reads stories, generates test cases, writes Playwright specs
• 🔌 MCP (Model Context Protocol) — connects AI to Jira, GitHub, TestRail automatically
• 📚 RAG (Retrieval Augmented Generation) — gives AI your private codebase knowledge

No fluff. No theory. Just working code and a clear learning path.

ai-qa-learning-roadmap/
│
├── 📄 README.md                  ← You are here
├── 🐍 orchestrator.py            ← Working AI QA pipeline (start here)
├── 📁 examples/
│   └── STORY-1042.spec.ts        ← Example Playwright output generated by AI
├── 📁 docs/
│   ├── architecture.png          ← Pipeline architecture diagram
│   ├── ROADMAP.md                ← 5-week learning roadmap
│   └── HOW_IT_WORKS.md           ← Plain English explanation of AI + RAG + MCP
└── 📁 coming-soon/
    ├── jira-mcp-connector.py     ← Stage 2 — Jira MCP (coming this week)
    └── rag-layer.py              ← Stage 3 — RAG with pgvector (coming soon)

# 1. Clone the repo
git clone https://github.com/duveyvaishnavi-stack/ai-qa-learning-roadmap

# 2. Install dependencies
pip install anthropic

# 3. Set your API key (get it free at console.anthropic.com)
export ANTHROPIC_API_KEY=sk-ant-your-key-here   # Mac/Linux
set ANTHROPIC_API_KEY=sk-ant-your-key-here       # Windows

# 4. Run the pipeline
python orchestrator.py

# Output: STORY-1042.spec.ts — ready-to-run Playwright file ✅

┌─────────────────────────────────────────────────┐
│              MCP Tools Layer                     │
│   Jira · GitHub · TestRail · Slack              │
└──────────────────┬──────────────────────────────┘
                   ↓
┌─────────────────────────────────────────────────┐
│            AI Orchestrator                       │
│  Claude reads stories · generates tests          │
│  writes Playwright specs · opens PRs             │
└──────────────────┬──────────────────────────────┘
                   ↓
┌─────────────────────────────────────────────────┐
│           RAG Knowledge Layer                    │
│  Your codebase · standards · past incidents      │
│  Embedded in pgvector (free)                     │
└─────────────────────────────────────────────────┘

💡 Best way to learn: Follow this repo — each stage ships as working code you can run and study.

• Reads user stories
• Generates happy path + edge cases + negative tests
• Writes ready-to-run Playwright TypeScript spec files

• Jira MCP — auto-fetch stories when they move to "Ready for QA"
• GitHub MCP — auto-open PRs with generated spec files
• Zero keyboard interaction between story and PR

• Embed your codebase, page objects, and standards
• AI generates tests using YOUR selectors and conventions
• Tests look like a senior QA on your team wrote them

"This is the kind of AI-in-QA content that feels practical instead of theoretical."

"The interesting part is not the 60 lines of Python. It's the orchestration thinking behind it."

"Where is the MCP and RAG layer?" — Stage 2 dropping this week! ⬆️ Star to get notified

Found a bug? Want to add a connector? PRs welcome!

1. Fork the repo
2. Create your branch: git checkout -b feature/jira-mcp
3. Commit your changes
4. Open a PR — I review within 24 hours

Vaishnavi Duvey — QA Manager & Business Analyst | AI-Assisted Testing & Automation

I share practical AI + QA content every week on LinkedIn.

🔗 LinkedIn · ⭐ Star this repo

Give it a star — it helps more QA engineers find this resource!

⭐ Star this repo — it helps more QA engineers find this!

Built by Vaishnavi