Project: Zapuza Talent Lab – The Enterprise-Grade Secure Assessment Ecosystem

Developed by Zapuza

Zapuza Talent Lab was engineered to solve the most complex challenge in remote evaluations: establishing a physical exam center’s integrity within a remote digital environment. We built a high-performance Secure Assessment Wrapper that encapsulates third-party exam links, creating an impenetrable digital fortress around the candidate.

Unlike standard browser extensions, this is a system-level application that ensures “Zero-Trust” security. It allows organizations to host exams on their own servers (maintaining full data ownership) while relying on Zapuza’s AI to handle security, surveillance, and evidence generation.

Key Features & Functionalities

1. “Iron-Clad” Kiosk Environment (System-Level Control)

• Total OS Lockdown: Upon launch, the wrapper takes control of the operating system, forcibly terminating prohibited background processes such as TeamViewer, AnyDesk, Discord, and screen-sharing tools.

• Peripheral & Hardware Control: Automatically disables secondary monitors, blocks clipboard functions (Copy/Paste), and prevents keyboard shortcuts (Alt+Tab, Win+D) to ensure the candidate cannot leave the assessment interface.

• Virtual Machine Detection: Advanced algorithms detect if the exam is being attempted from inside a Virtual Machine (VM) or Sandbox to bypass security, blocking access immediately.

2. 360-Degree Multi-Sensory Monitoring

• Tri-Channel Surveillance: The system simultaneously monitors and synchronizes three distinct data streams:

  • Webcam Feed: For identity verification and behavioral tracking.

  • Microphone Audio: To detect unauthorized conversations or background coaching.

  • Desktop Screen: Continuous recording of the user’s screen activity to ensure no on-screen cheating occurs.

• Audio Environment Analysis: AI monitors ambient noise levels, flagging sudden spikes in volume or detecting human speech patterns distinct from the candidate.

3. Advanced AI Anomaly Detection (Computer Vision)

• Real-Time Object Detection: The AI is trained to recognize unauthorized objects within the frame. It instantly flags:

  • Mobile Phones / Tablets

  • Books / Notes

  • Additional Electronic Devices

• Biometric Integrity:

  • Face Tracking: Ensures the registered candidate is present 100% of the time.

  • Multiple Face Detection: Flags the session immediately if a second person enters the camera frame.

  • Absence Detection: Logs an incident if the candidate leaves the seat or covers the camera.

4. Forensic-Grade Reporting & Evidence Logs

• Visual Evidence Snapshots: The system does not just flag an error; it captures proof. Every time an anomaly is detected (e.g., a phone is seen), a high-resolution screenshot is taken and embedded into the final report.

• Focus Loss & Navigation Logs: Detailed tracking of “Window Focus” events. Even if a user attempts to minimize the wrapper or switch tabs, the exact timestamp and duration of the attempt are recorded.

• The “Black Box” HTML Report: At the end of the session, a standalone, encrypted HTML report is generated and uploaded to the client server. This report includes:

  • Second-by-second activity logs.

  • Embedded images of every flagged incident.

  • Audio anomaly timestamps.

  • A summary “Trust Score” for quick evaluation.

5. Privacy-First “Zero-Touch” Architecture

• Client-Server Isolation: The candidate logs in using the client’s own credentials on the client’s server page loaded inside our wrapper. Zapuza never accesses the question bank or user passwords.

• Secure Data Pipeline: All proctoring data (logs, images, reports) is processed locally or in a secure temporary state and then pushed directly to the client’s storage bucket, ensuring we do not retain sensitive user data post-exam.

Business Impact & Benefits

• Uncompromised Integrity: By combining system lockdown with tri-channel monitoring (Video/Audio/Screen), we eliminated 99% of common remote cheating methods.

• Verifiable Evidence: The inclusion of “Evidence Snapshots” in the HTML report allows administrators to validate cheating claims definitively, reducing disputes.

• Universal Wrapper Utility: The solution is platform-agnostic. Whether the client uses Moodle, a custom PHP portal, or a legacy LMS, Zapuza Talent Lab secures it instantly without API rework.

• Scalable Architecture: Capable of handling thousands of concurrent test-takers with lightweight, optimized AI models running efficiently on the client side.

Conclusion

Zapuza Talent Lab is more than just a proctoring tool; it is a comprehensive exam security infrastructure. By fusing object detection, behavioral analysis, and system-level restriction into a seamless wrapper, we have empowered institutions to conduct high-stakes examinations with the confidence of a physical invigilator. This project stands as a benchmark for secure, privacy-conscious EdTech innovation.