AI-Powered Business Assistant (Multi-Tenant RAG Platform)
Designed and built a multi-tenant AI SaaS platform that allows organizations to securely upload internal documents and query them using Retrieval-Augmented Generation (RAG). Focused on reducing hallucinations while delivering accurate, context-aware responses at scale.
Implemented intelligent document chunking, vector embeddings, semantic retrieval, strict prompt constraints, and organization-level data isolation with role-based access control.
Tech: Node.js, PostgreSQL, pgvector, React Native, LLMs, RAG