Local LLM Test Generator
Full-stack chat app that turns pasted Jira user stories into structured test cases. Works with Claude, OpenAI, Groq, Gemini, and fully local Ollama or LM Studio models.
RICE-POT Framework Generator
Prompt engineering as a power tool: one RICE-POT master prompt generates an enterprise Selenium TestNG POM framework, a Playwright suite, and a full VWO test plan.
JobTracker AI
Local-first Kanban board for the QA job hunt: drag applications from Wishlist to Offer, everything stored in your browser, plus an Easy-Apply agent playbook.
Test Planner Agent
Point it at any Jira or Azure DevOps ticket and it drafts a complete, template-driven test plan document, built with the B.L.A.S.T. agent framework and deployed on Vercel.
n8n QA Agent Fleet
Six no-code agents: a QA expert chatbot, a local Ollama variant, a Jira bug creator, a PRD-to-test-plan reader, and PRD-to-test-cases flows that file Jira subtasks or export Excel.
LangFlow QA Buddy
Visual drag-and-drop agents: a QA Buddy chat assistant and a bug-report classifier that assigns category and priority, both as importable LangFlow canvases.
AI Test Case Generator
VIBE-coded in a week: FastAPI + React app that pulls a Jira story, applies a YAML template (smoke, regression, edge, security), and returns editable, exportable test cases.
RAG Explained: Basic
A glass-box RAG pipeline over the VWO PRD: watch the PDF get chunked, embedded, and stored in ChromaDB, then see exactly which chunks answer your question and why.
Advanced RAG Lab
Hybrid retrieval over 5,000 VWO test cases: BGE-M3 dense + sparse vectors in Qdrant, RRF fusion, cross-encoder reranking, query rewriting, and a stage-by-stage inspector UI.
QA Copilot
The RAG-track capstone: one chat box grounded in five knowledge sources, Selenium code, Playwright code, test cases, PRDs, and JIRA bugs, with cited answers and a RAG Explorer.
Playwright MCP + First Server
Let an AI drive a real browser through Playwright MCP (23 tools) to test the VWO login and auto-write the spec file, then author your own FastMCP server from scratch.
Python for Testers
Twelve runnable lessons that take a manual tester to Python fluency: variables to tuples, plus calling an LLM from pure Python on lesson four.
TestCase MCP Server
A FastMCP server that turns a 478-row VWO test-case CSV into 15 tools, 3 resources, and 2 prompts, so Claude Desktop or Cursor can search, filter, and even add test cases.
CrewAI Bug Triage Crew
Three specialists on one bug: a Triage Analyst sets severity, a Root Cause Specialist digs into the why, and a Test Strategy Advisor writes the Playwright prevention tests.
CrewAI QA Pipeline
Give it one Jira ticket ID and four agents hand work down the line: read the ticket over MCP, write the test plan, design the cases, and code the Playwright tests.
Production QA Pipeline
The Chapter 14 crew, productionized: templated plans, Jira-importable CSVs, a generated multi-file Playwright framework, and a web UI that fans out over many tickets.
LLM-as-Judge with DeepEval
Stop eyeballing AI output. Score answer relevancy and hallucination with DeepEval, then put Llama-4 on trial with GPT-4.1 as the judge, via Groq and OpenRouter.
Fine-Tuning Open Models
Adapt Qwen2.5-Coder to your private QA data: when RAG is enough, when LoRA/QLoRA wins, and a full local recipe with Ollama, LanceDB, and MLX.
DeepEval Framework
A production evaluation harness: 29 metric-target pairs across a live e-commerce chatbot, a RAG pipeline, and a black-box browser bot, with a 4-panel metrics dashboard.
BrowserBash NL E2E
Write the journey in plain English, no selectors: login, add to cart, checkout. An AI agent drives the real browser and returns a step-by-step pass/fail report.
JIRA Bug Triage Agent
A LangChain agent that reads raw Jira bugs and returns structured triage: severity, priority, component, owning team, and likely duplicates, grounded in a RAG knowledge base.
RCA Agent
Feed it a failure and it investigates like an SDET: pulls Jenkins console logs, correlates recent GitHub commits, pings the live endpoint, and writes the root-cause verdict.
Test Designer Agent
From user story to test plan in one run: retrieves similar existing cases from the RAG knowledge base so it never duplicates, then designs a structured TestPlan (18 cases for VWO-105).
QA Copilot Router
One box, three brains: a router classifies each request and dispatches to the Triage, RCA, or Test Designer agent, unified behind a single Streamlit UI with graceful fallbacks.
AI Testing Agents Studio
Thirteen portfolio projects, 10 agents and 3 tools, each shipped three ways: importable n8n flow, importable LangFlow canvas, and a React Studio that runs them through 6 LLM providers.