Built a full-stack QMS application using a 7-node LangGraph agent pipeline (Groq LLMs) to auto-extract complaint data, classify risk, detect duplicates, and recommend CAPA from uploaded pharma complaint documents • Implemented real-time auto-save and chat-driven form editing with React + Redux Toolkit; FastAPI + SQLAlchemy backend supporting multiple databases

Built an n8n-style visual AI workflow builder with LLM, HTTP, database, conditional, notification, and approval nodes. • Implemented manual, webhook, scheduled, and database-event triggers with live execution status. • Added Nhost Auth, Hasura RBAC, organization management, and approval-based workflow pause/resume. • Integrated Groq API for real-time LLM execution and designed a modular Node.js/Express backend.