Meta Llama 4 Hackathon 1st Place Winning Project - https://x.com/MetaforDevs/status/1937233386453762423

A comprehensive classroom monitoring solution that uses AI to analyze student screen activity in real-time. The system consists of a macOS student client app and a web-based teacher dashboard, connected through MongoDB for seamless data synchronization.

Sussi combines real-time screen capture, AI analysis, and intuitive teacher tools to create a modern classroom monitoring experience. Teachers can set assignments, monitor student progress, and receive intelligent insights about student engagement and focus levels.

• Real-time screen capture using ScreenCaptureKit (every 10 seconds)
• AI-powered analysis using Llama-4-Scout-17B-16E-Instruct-FP8
• Automatic assignment sync from MongoDB (every 5 seconds)
• Focus score tracking with intelligent scoring algorithm
• Screenshot storage with base64 encoding for teacher review

• Real-time student monitoring with live activity updates
• Assignment management with instant distribution to all students
• AI chat assistant (Sussi AI) for classroom insights
• Student status indicators (On-Task, Suspicious, Needs Help)
• Multiple view modes (Grid and Heatmap)
• Automation rules for proactive student management

• Centralized data storage for students, assignments, and messages
• Real-time synchronization between student apps and teacher dashboard
• Scalable classroom management with multi-class support

• Intelligent Screen Analysis: 4-field AI analysis including focus score (0-5), activity description, 3-word summary, and teacher suggestions
• Privacy-First Design: Local processing with secure API calls, no permanent screenshot storage
• Dynamic Focus Scoring: Adaptive algorithm that tracks student engagement over time
• Window Detection: Context-aware analysis considering active applications
• Assignment Integration: Automatic retrieval and display of teacher assignments

• Color-coded status indicators: Green (on-task), Orange (suspicious), Red (needs help)
• Real-time feedback: Live display of AI analysis and current assignment
• Minimal system impact: Efficient screen capture with optimized API calls

• Live Activity Grid: Real-time view of all student screens and activities
• Individual Student Cards: Detailed view with screenshots, status, and activity history
• Student Chat: Direct communication with Sussi AI about specific students
• Classroom Chat: AI assistant for overall classroom management

• Dynamic Assignment Distribution: Set and update assignments instantly across all students
• Flag Summary: Automatic detection of students not following instructions
• Progress Monitoring: Real-time updates on student compliance and engagement

• Automation Rules: Set up triggers for common classroom scenarios
• Multiple View Modes: Switch between detailed grid and classroom heatmap views
• Message System: Send targeted messages to individual students
• Mock Data Support: Built-in demo mode for testing and training

• Platform: macOS 13.0+ (Swift/SwiftUI)
• Screen Capture: ScreenCaptureKit framework
• Database: MongoDB with MongoSwiftSync driver
• AI Integration: Direct API calls to Llama service
• Image Processing: Base64 encoding with 720p optimization

• Framework: Next.js 14 with App Router
• Language: TypeScript
• Styling: Tailwind CSS v4
• Database: MongoDB with official Node.js driver
• State Management: React hooks
• UI Components: Custom components with modern design

• macOS 13.0+ (for student client)
• Node.js 18+ (for teacher dashboard)
• MongoDB Atlas account or local MongoDB instance

1. Configure MongoDB connection:

   cd LlamaStudentClient
   cp Config.template.swift LlamaStudentClient/Config.swift
   # Edit Config.swift with your MongoDB connection string

2. Build and run:

   • Open LlamaStudentClient.xcodeproj in Xcode
   • Build and run the project (⌘+R)
   • Grant screen recording permissions when prompted

1. Install dependencies:

   cd teacher
   npm install

2. Configure environment:

   # Create .env.local file with:
   MONGODB_URI=your_mongodb_connection_string
   LLAMA_API_KEY=your_llama_api_key

3. Run development server:

   npm run dev

   Navigate to http://localhost:3000

Create the following collections in MongoDB:

• LlamaProctorDB.students - Student activity data
• LlamaProctorDB.assignments - Teacher assignments
• LlamaProctorDB.messages - Student-teacher messages

1. Assignment Creation: Teachers set assignments through the web dashboard
2. Assignment Sync: Student apps retrieve assignments every 5 seconds
3. Screen Capture: Student apps capture screenshots every 10 seconds
4. AI Analysis: Screenshots analyzed by Llama AI for activity assessment
5. Data Storage: Analysis results stored in MongoDB with focus scores
6. Teacher Updates: Dashboard displays real-time student activity and insights

• Local Processing: Screenshots analyzed via API, not stored permanently on devices
• Secure Credentials: MongoDB URI and API keys stored in gitignored config files
• Permission-Based: Requires explicit screen recording permission from students
• Encrypted Transport: All data transmission uses HTTPS/TLS encryption
• Masked Logging: Sensitive credentials automatically masked in console output

• Classroom Management: Real-time monitoring of student engagement and focus
• Remote Learning: Ensure students stay on task during online classes
• Assessment Integrity: Monitor student activities during digital assessments
• Behavior Analytics: Track patterns in student engagement over time
• Intervention Alerts: Automated notifications when students need assistance

• Multi-platform Support: Windows and Linux student clients
• Advanced Analytics: Detailed engagement reports and trends
• Integration APIs: Connect with popular LMS platforms
• Mobile Dashboard: iOS/Android apps for teachers
• Voice Commands: Hands-free classroom management
• Parent Portal: Optional parent access to student progress

This project is part of an educational monitoring system designed for classroom use. Please ensure compliance with local privacy laws and institutional policies before deployment.