A self-hosted, multi-source Hybrid RAG system giving QA engineers one cited answer grounded across Selenium/Playwright framework repos, a 5,000+ test case repository, JIRA tickets, PRDs/BRDs/SRSs, meeting notes, and Jenkins logs — with metadata-aware retrieval, hourly auto-ingestion with deletion handling, and an MCP server exposing the assistant as tools for team-wide use.

RAG Qdrant Groq FastAPI MCP Hybrid Search Reranking DeepEval Docker

An advanced RAG pipeline built in Langflow to retrieve from a repository of 5,000+ test cases — combining HyDE query generation, semantic search (ChromaDB + Mistral embeddings), NVIDIA reranking, parent-document retrieval for full test-case context, and contextual compression before generating grounded answers with Groq.

Langflow ChromaDB Mistral Embeddings NVIDIA Rerank HyDE Groq RAG RAGAS

A Retrieval-Augmented Generation assistant built in Langflow that searches an e-commerce test-case repository and returns grounded, natural-language answers — using ChromaDB for vector storage, Mistral embeddings, and Groq's Llama 3.1 8B for generation. Includes a dedicated test suite covering positive retrieval, hallucination prevention, and prompt-injection resistance.

Langflow ChromaDB Mistral Embeddings Groq RAG

Production-ready n8n AI agent that reviews Jira stories for QA refinement readiness — fetches ticket details, generates a structured 14-section QA report (readiness score, gaps, test scenarios, clarification questions), and posts it straight to Slack. Includes a Promptfoo evaluation suite for regression-testing the agent's scoring consistency.

n8n OpenAI Jira API Slack API Promptfoo AI Agents

A deployed React frontend (Vite, hosted on Vercel) for a document-upload-and-chat interface powered by an n8n RAG pipeline — built as a public, end-to-end portfolio piece demonstrating real deployment work (env var config, iframe integration, production build troubleshooting).

React Vite Vercel n8n RAG

A two-phase Retrieval-Augmented Generation system built in n8n — document ingestion (chunking, embedding, vector storage) and a conversational retrieval agent with chat memory, powered by Groq's Llama 3.1 8B and Pinecone.

n8n RAG Pinecone Groq OpenAI Embeddings

A multi-agent n8n workflow that classifies IT support tickets, auto-drafts responses for low-risk requests, and routes uncertain or high-impact tickets to a human via Slack for approval — with full observability logging to Google Sheets.

n8n Groq AI Agents Human-in-the-Loop

Pulls requirements from Jira and uses the Anthropic API to auto-generate structured test cases — reduces manual test design effort by 70–80%.

Python Anthropic API Jira API

End-to-end UI automation framework using Page Object Model — reduces testing cycles by ~70%.

Playwright TypeScript

E-commerce UI automation suite covering login, cart, and checkout flows.

Playwright TypeScript