Back to All Projects
Problem Statement
Remote technical screening suffers from high proctoring overhead and manual resume screening biases.
Proposed Solution
Integrated PyTorch computer vision tracking with automated resume parsing and candidate ranking.
Key Features
- Automated PDF & DOCX Resume Parsing with OCR extraction
- Real-time webcam head pose & eye gaze gaze tracking proctoring
- AI interview simulator with automated scoring engine
- Recruiter Dashboard with exportable candidate analytics
System Architecture Flow
Next.js Client -> Python OpenCV/PyTorch Service -> MERN Backend Database
Engineering Challenges
Optimizing real-time gaze detection FPS without taxing candidate web browser hardware.
Lessons Learned
Fine-tuning computer vision model thresholds for browser WebRTC streams.