Secure Online Exams,
built as a team.
A fully proctored examination portal: biometric face verification, real-time webcam monitoring for cellphones and missing or multiple faces, and AI grading for short-answer questions. Built with Antigravity — and, more importantly, built with two classmates, which is where I actually learned team management and problem solving.
Visuals generated with AI
All images below are AI-generated illustrations of the build
The student view keeps the question, the timer, the warning count and the live proctoring feed in one frame, so nothing about the rules is hidden.
A face profile registered on first login is matched against a live snapshot before a camera-enabled exam can start.
Question building for MCQ, true/false and short answer, Gemini-assisted grading of subjective responses, and a timestamped proctoring log for review.
Grade distribution, flag distribution and submission timelines, plus user management and downloadable PDF performance reports.
Three people, one system
Remote exams only work if everyone trusts them. The brief we set ourselves was to make integrity checks feel like part of the exam rather than surveillance bolted on top: verify once, warn clearly, log everything a faculty member might need to review later.
Under the hood it is a Flask application with OpenCV doing face tracking and MobileNetSSD cellphone detection, Gemini grading short-answer responses against the faculty key, and ReportLab producing the admin PDF reports. Antigravity carried a lot of the implementation; the decisions about what belonged in the product were ours.
I built this with two classmates, Chaitanya and Meghana, who helped enormously. Splitting the work, unblocking each other and arguing about edge cases taught me more about team management and problem solving than the code did.
Note: the interface visuals on this page are AI-generated illustrations created to represent the system, not screenshots of the running application.