Back to Projects
AI & Streaming• Production Verified
AI Exam Proctoring Service
Real-time online exam monitoring service with AI-based anomaly detection for Learning Management Systems.
Key Verified Outcomes
Real-time AI
Cheating Detection
< 150ms
Latency Overhead
Seamless
LMS Integration
Live System Walkthrough & Demonstration
Visual walkthrough of core application features, system architecture, and performance.
2:40 min live test
0:003:00
The Architectural Challenge
Developed a robust exam proctoring service integrated into the Learning Management System (LMS) at Darvilab. The system monitors remote test-takers in real time and uses AI to flag suspicious activities and prevent cheating.
Engineering Execution & Deep Dive
Building remote proctoring requires balancing candidate privacy, network constraints, and detection accuracy. We integrated lightweight computer vision and audio analysis streams with real-time WebSocket signals. The service runs unobtrusively in modern browsers and alerts proctors immediately when anomalies are identified.
Core Technical Features:
- Real-time video and audio stream verification for remote exam sessions.
- AI-based detection for suspicious head movement, multiple faces, and unauthorized browser tabs.
- Secure proctor dashboard providing synchronized video playback and incident timelines.
- Automated session logging with end-to-end encrypted storage on AWS S3.
Security Hardening & VAPT
- ✦Encrypted WebRTC streams preventing man-in-the-middle packet tampering.
- ✦Role-based access allowing only verified educators and proctors to review exam audit trails.
- ✦Secure token-based session initialization preventing exam credential sharing.
Technology Stack
ASP.NET CoreC#ReactNode.jsAI ServicesWebSocketsWebRTCAWS EC2 & S3PostgreSQL
Engineering Governance
This project follows strict code review guidelines, automated static code analysis, and regression verification before production deployment.