Back to projects
/ AISDOP product · AI-monitored aptitude tests
Product in developmentAI-MAT
An online aptitude-testing product that combines an exam journey with browser-based AI-assisted monitoring and proctoring signals.
01 / CONTEXT
The problem
Online aptitude assessments need a clear candidate journey and tools to surface possible integrity concerns during a test.
02 / DIRECTION
The approach
Structure the test journey around consent, room scan, identity capture, assessment, and results, while using client-side models to detect selected visual signals and record review events.
03 / SYSTEM MAP
Architecture at a glance
01Pre-exam consent and environment checks
02Identity capture flow
03Assessment and answer state
04TensorFlow.js models in a Web Worker
05Results and proctoring event summary
04 / ENGINEERING NOTES
What this explores
Product of AISDOP
Uses COCO-SSD and BlazeFace according to the repository documentation
Separates AI inference from the main UI thread
AISDOP Compiler is part of the wider product ecosystem
Replace or supplement this summary with verified implementation details, testing evidence, screenshots, and your personal contributions before presenting the work as production-complete.