N.
All work

QuerioAI Document Intelligence Platform

04 / 07
04
Year
2025
Role
Full-Stack AI Developer
Duration
2 months
Client
Vennisys

Live project Source

Overview

Built a modern full-stack application that leverages Retrieval-Augmented Generation to enable natural-language conversations with documents. Users upload PDFs, ask questions in plain English and receive intelligent, context-aware answers.

Challenge

Organizations struggle to extract meaningful insight from large document collections. Traditional keyword search falls short for complex queries that require context and semantic understanding.

Solution

Engineered an async FastAPI backend integrating Google Gemini Pro with a ChromaDB vector database and HuggingFace embeddings. Built a modern Next.js frontend with drag-and-drop upload and session-based conversational AI with context retention.

Results

  • 01

    40–60ms inference latency per query

  • 02

    Multi-session conversational AI with context retention

  • 03

    Real-time semantic search across documents

  • 04

    Zero-setup document onboarding via drag-and-drop

Technology

Next.js 16React 19TypeScriptPythonFastAPILangChainGoogle Gemini ProChromaDBHuggingFaceTailwindCSSZustand