System Requirements Document (SRD) for Agile-Download

1. Introduction

The Agile-Download project is envisioned as an AI-powered online exam cheating detection system designed to ensure the integrity of remote assessments. By leveraging cutting-edge technologies such as computer vision, machine learning, and real-time monitoring, this system will detect and prevent cheating during online exams. The solution will cater to academic institutions, corporate organizations, and certification bodies, ensuring fairness, security, and a seamless user experience.

This document outlines the system requirements for Agile-Download, focusing on its functional and non-functional aspects, user personas, visual design, and technical architecture.

2. System Overview

Agile-Download is a comprehensive AI-powered online exam cheating detection system that integrates artificial intelligence and machine learning to detect and prevent cheating in real-time. The system will utilize advanced technologies such as:

• Computer Vision for face detection and eye tracking.
• Behavior Analysis to identify suspicious activities.
• Webcam and Audio Surveillance for real-time monitoring.
• Tab Switching Detection to prevent unauthorized access during exams.
• Impersonation Detection to ensure the authenticity of test-takers.
• Data Security and Authentication to protect sensitive information.
• API Supervisor Integration for pre-resolved cheating detection and system monitoring.

The system will be accessible via a web-based interface and a mobile application, ensuring compatibility across devices and platforms. It will support both automated and live proctoring options, catering to diverse user needs.

3. Functional Requirements

The functional requirements for Agile-Download are outlined as user stories:

• As a User, I should be able to log in securely using multi-factor authentication.
• As a User, I should be able to take an online exam with real-time monitoring.
• As the System, I should detect and flag instances of tab switching during an exam.
• As the System, I should perform face detection to verify the identity of the test-taker.
• As the System, I should track eye movements to detect potential cheating behavior.
• As the System, I should analyze user behavior to identify suspicious activities.
• As the System, I should monitor audio input to detect unauthorized communication.
• As the System, I should provide real-time alerts for impersonation detection.
• As the System, I should integrate with an API supervisor to pre-resolve potential cheating incidents.
• As an Admin, I should be able to review flagged incidents with detailed reports.
• As an Admin, I should be able to configure exam settings, including time limits and allowed resources.
• As an Admin, I should be able to manage user accounts and permissions.
• As an Admin, I should be able to download and export exam results securely.

4. User Personas

4.1 Test-Taker (User)

• Description: A student or professional taking an online exam.
• Goals: Complete the exam without interruptions, ensure compliance with proctoring rules.
• Pain Points: Privacy concerns, technical issues during exams.

4.2 Administrator (Admin)

• Description: An academic or corporate professional managing the examination process.
• Goals: Ensure exam integrity, review flagged incidents, manage user accounts.
• Pain Points: Complexity in managing large-scale exams, false positives in cheating detection.

4.3 Proctor (Optional Role)

• Description: A human proctor overseeing live exams.
• Goals: Monitor test-takers in real-time, intervene when necessary.
• Pain Points: Limited visibility into user activities, reliance on automated alerts.

4.4 API Supervisor

• Description: A backend AI system that pre-resolves potential cheating incidents by analyzing data in real-time.
• Goals: Enhance system efficiency by reducing false positives and providing actionable insights to administrators.
• Pain Points: Requires high computational resources and robust integration with the system.

5. Visuals Colors and Theme

The visual design of Agile-Download will reflect professionalism, trust, and technological sophistication. The following color palette has been designed specifically for this project:

• Background: #EAF2F8 (Soft Ice Blue)
• Surface: #FFFFFF (Pure White)
• Text: #1C2833 (Midnight Black)
• Accent: #5DADE2 (Bright Azure Blue)
• Muted: #ABB2B9 (Steel Gray)

These colors will create a clean and modern interface, ensuring readability and ease of use.

6. Signature Design Concept

The home page of Agile-Download will feature an immersive AI-powered visualization that highlights the system's capabilities. The design concept includes:

• Interactive Cheating Detection Demo: A live simulation where users can see how the system detects cheating in real-time, with visual overlays for face detection, eye tracking, and tab switching.
• AI Supervisor Dashboard: A dynamic display showcasing the API supervisor's real-time analysis and decision-making process, with animated graphs and alerts.
• Behavior Analysis Heatmap: A heatmap animation that demonstrates how the system identifies suspicious activities during exams.
• Seamless Transitions: Smooth animations between sections, with micro-interactions like hover effects and button clicks.
• AI Persona Animation: A friendly AI avatar that guides users through the system's features, adding a human touch to the interface.

This bold and engaging design will make a lasting impression on users, emphasizing the system's advanced capabilities and user-centric approach.

7. Non-Functional Requirements

• Performance: The system must process real-time video and audio streams with minimal latency.
• Scalability: The system should support up to 10,000 concurrent users during peak exam periods.
• Security: All data must be encrypted in transit and at rest, adhering to GDPR and other relevant regulations.
• Availability: The system must maintain 99.9% uptime during exam periods.
• Usability: The interface must be intuitive and accessible, with support for multiple languages.
• Compatibility: The system should work on major browsers (Chrome, Firefox, Edge) and mobile platforms (iOS, Android).
• Integration: The system must seamlessly integrate with the API supervisor for pre-resolved cheating detection.

8. Tech Stack

The following technologies will be used to build Agile-Download:

• Frontend: React for web, React Native for mobile.
• Backend: Python with FastAPI.
• Database: MySQL for relational data, MongoDB for NoSQL data.
• AI Models:
  • GPT 5.4 for user-friendly responses.
  • Claude 4.6 Opas for academic or coding work.
  • Google Nano Banana for image generation.
• AI Tools: Litellm for LLM routing, Langchain for orchestration.
• Orchestration: Docker for local, Kubernetes for server-side.
• API Supervisor: Integrated for pre-resolved cheating detection and system monitoring.

9. Assumptions and Constraints

• The system assumes a stable internet connection for real-time monitoring.
• Users must grant access to their webcam and microphone for the system to function.
• The system will not store raw video or audio data to ensure privacy.
• The initial release will focus on English, with support for additional languages in future updates.
• The API supervisor requires robust computational resources for real-time analysis.

10. Glossary

• AI (Artificial Intelligence): The simulation of human intelligence in machines.
• ML (Machine Learning): A subset of AI focused on building systems that learn from data.
• Computer Vision: A field of AI that enables machines to interpret visual information.
• Proctoring: The process of supervising an examination.
• Tab Switching Detection: A feature that identifies when a user switches away from the exam interface.
• Impersonation Detection: A feature that verifies the identity of the test-taker to prevent fraud.
• API Supervisor: A backend AI system that pre-resolves potential cheating incidents by analyzing data in real-time.