✅ VerifyHub: AI-Powered Document & Talent Show Verifier

VerifyHub is a powerful, modern web application built to streamline and automate the verification of student academic documents and talent show submissions. With integrated AI, web agents, and document processing tools, it aims to reduce human error, speed up evaluations, and ensure originality with minimal manual effort.

🚀 Inspiration

Manual verification of student transcripts and creative submissions (like lyrics or scripts) is often tedious, slow, and error-prone. This leads to bottlenecks in college applications, talent shows, and similar events.

We built VerifyHub to solve this by leveraging advanced AI tools like Gemini AI, Vertex AI, Document AI, and agent-based reasoning using CrewAI with Python. The result is a fast, secure, and scalable solution for intelligent document verification.

💡 What It Does

🧾 Student Document Verification

• Users upload academic documents (e.g. transcripts in PDF format).
• The system extracts and analyzes the content using Document AI.
• An AI model verifies the formatting, content, and integrity.
• A “VERIFIED” stamp with college logo is added to the document.
• The stamped PDF is downloadable for official submissions.

🎤 Talent Show Submission Verification

• Users upload lyrics, scripts, or performance descriptions (PDF/txt).
• The system uses CrewAI agents + Gemini AI + web search tools to:
  • Search the internet for similar existing content.
  • Assess originality and plagiarism likelihood.
  • Provide an AI-generated originality score and explanation.

🧠 AI-Powered Core

• Gemini AI for language-based feedback and analysis.
• Vertex AI for classification, fine-tuning, and advanced validations.
• Google Document AI for extracting structured data from uploaded documents.
• CrewAI + Python for orchestrating agent-based web browsing and checking prior submissions.

🛠️ How We Built It

👨‍💻 Frontend

• React + Vite for a high-performance web app.
• Material-UI (MUI) for modern, accessible UI components.
• Framer Motion for slick animations and transitions.
• Responsive, dark-mode optimized layout with smooth scrolling.

📄 PDF Handling

• pdf-lib to modify, stamp, and save PDFs.
• Custom logic for injecting college logo and “Verified” text.
• canvas-confetti for fun celebration effects after verification.

🤖 AI + Backend Agents

• @google/generative-ai: Gemini integration for NLP tasks.
• Google Document AI: Content extraction from PDFs.
• Google Vertex AI: Custom models trained on past student data.
• CrewAI + Python Agents:
  • Browser tool agents search public records or forums.
  • Determine if a talent submission is original or copied.
  • Feedback is relayed back to the UI with explanation.

🔍 Architecture Overview

| User Uploads PDF/Text |
           ↓
| React Frontend (UI + PDF Upload) |
           ↓
| AI API Layer (Gemini / Vertex AI / Document AI) |
           ↓
| CrewAI Python Agent Runner (Web Browser Tool Access) |
           ↓
| Verification Result + Stamped PDF + Originality Score |
           ↓
| User Downloads Verified Output |

## Challenges we Faced
- Handling large PDF files efficiently in the browser.
- Injecting high-resolution images (logos) into PDFs with pdf-lib.
- Orchestrating 3 different Google AI tools cohesively.
- Fine tuning Vertex AI on past transcripts

## Accomplishments
- Fully functional client-side document verification and stamping.
- Originality checker powered by CrewAI and web search tools.
- Smooth integration of Gemini + Vertex + Document AI.

## What We Learned
- How to use Google Cloud Platform (GCP) tools like Vertex AI, Document AI, and Generative AI effectively.
- Fine-tuning AI models using Vertex AI and integrating them with React-based web applications.
- Using CrewAI for dynamic web agent orchestration.

Built With

• document-ai
• gemini
• react
• vertex