Summary: The selected candidate will develop a deep learning-based crack detection algorithm for Company's BrakeView DiscAxle and BrakeView PadAxle systems. The project involves creating a PyTorch-based machine vision algorithm (MVA) to identify cracks in brake disk images, optimized for edge deployment.
Key Responsibilities:
Design and implement a computer vision algorithm for disk crack detection using PyTorch
Train and optimize deep learning models to achieve 95% accuracy
Optimize models for edge deployment (≤150MB size, ≤100ms inference)
Validate models using Company-provided datasets
Conduct UAT testing on production simulators
Participate in weekly code reviews and knowledge transfer sessions
Required Skills:
7/8+ years of experience with Python and deep learning frameworks
Strong PyTorch experience (mandatory - not TensorFlow)
Computer vision experience with defect/anomaly detection
Experience optimizing models for edge deployment
Understanding of object detection metrics (mAP, IoU, precision/recall)
