AI proctoring for online exams

During an internship at Pheme Software, I built an AI-based online examination system with anti-cheating features and integrated it into LMS platforms used by schools and training providers.

The system used PyTesseract-based facial recognition to monitor student behavior during exams, reaching 95% accuracy on misconduct detection. Cheating incidents dropped 20% and user engagement rose 40% after rollout. There was also a recommendation layer for personalized study tips based on performance (+30% retention in early trials).

Focus

• AI proctoring integrated with LMS platforms
• Facial recognition for exam integrity
• End-to-end exam workflow, not just a detection model