AvoidCheating

Home Page with Problem Statement, Description, Solution and Key Features and Survey Form
Dashboard Page to Start the Interview
Proctoring Analytics Advanced insights into interview integrity
AI Detection Logs Events captured during the interview session.
Final Interview Integrity Report AI analysis of candidate behavior and cheating attempts.
Interview Features Including AI Procotor, Interview Notes, Coding Playground, Live Camera and Secure Chat box with Interviewer
Survey Form

AvoidCheating is an AI-powered platform designed to ensure fairness and integrity in remote technical interviews. With thousands of candidates using AI tools, taking external help, or switching applications during an interview, it has become difficult for recruiters to identify real talent. This project solves that problem by building a secure, intelligent, and cheat-proof interview environment.

Inspiration: The inspiration came from observing how remote interviews are easily compromised today: Candidates secretly using ChatGPT or other AI assistants Switching tabs to search for answers Taking help from a second person off-screen Using external devices to cheat Turning off the camera/mic to hide actions We wanted to build a platform where skills matter more than shortcuts, and interviewers can trust the results.

Problem We Solved: Every day, lakhs of remote interviews take place, and companies face challenges like: No reliable way to detect cheating Poor visibility into candidate behaviour No unified environment for code, camera, mic, and monitoring Lack of automated AI-powered proctoring Manual evaluation becomes lengthy and inaccurate AvoidCheating provides a single secure workspace with live proctoring using advanced AI and ML models.

How We Built the Platform: We built a fully integrated Interview Room with:

🔹 Coding Playground A full-featured code editor where candidates solve problems without switching tabs.

🔹 Camera & Mic Monitoring Constant face detection, gaze tracking, multiple-person detection, and audio anomaly detection.

🔹 AI Proctoring Engine The system monitors candidate behavior continuously and gives alerts for: Inactivity Eye-gaze anomalies No face detected Multiple faces detected Background noise App/tab switch Clipboard use Microphone mute Cheating patterns All alerts appear as 5-second toast messages.

🔹 Secure Lock-In Mode The candidate cannot exit full-screen or leave the interview until clicking End Interview.

🔹 Final Report Generation A clean summary report showing: Violations Timeline Risk score Session status

What We Learned:

AI/ML Integration Challenges Integrating vision-based detections (face, eyes, persons) with browser APIs required careful design to ensure accuracy and minimal false positives.
UX Matters in Proctoring Systems We learned that proctoring tools must be: clean simple responsive non-intrusive We applied toast notifications and toggles to avoid clutter.
Browser Limitations Browsers cannot block app switching completely. Instead, we learned to implement detection-based workflows which is the real industry standard.
Security First We learned how to design workflows to prevent: Page reload escape Back navigation Hidden tab switching Camera/mic disabling
How to Build a Unified Workspace Combining coding, video, audio, AI, and real-time alerts into one UI taught us a lot about UI/UX orchestration.

⚠️ Challenges Faced

Camera visibility & workflow conflicts We faced issues like camera not showing, initializing multiple times, overlapping workflows, and fixing alignment.
Accurate Eye & Face Detection Small movements triggered false warnings. We tuned thresholds for a smoother experience.
Layout alignment issues The interview room was getting messy due to multiple components.

We redesigned it into: Left: Coding Playground Top-right: Camera Bottom-right: AI tools in toggles

Preventing Escape Locking users inside fullscreen without breaking UX was challenging.
Risk Scoring Model Making sure different violations contributed correctly to the risk score required iteration.

AvoidCheating is more than just an interview tool — it is a complete AI-powered integrity system for remote hiring. The platform increases fairness, builds trust, and helps recruiters evaluate true candidate skill without external influence.

This project allowed us to combine: AI Frontend design Browser APIs Proctoring industry standards UX engineering…into one seamless experience.

What’s Next for AvoidCheating: AvoidCheating is just the beginning of a more secure and transparent remote hiring ecosystem. In the next phase, we plan to expand the platform with advanced capabilities that make interviews even more reliable, scalable, and AI-assisted. Optimizing the platform to support thousands of candidates simultaneously for college drives, exams, and hackathons with distributed AI processing through MVP.