AI Welding Seam Quality Inspection Developer

Ontix Intelligence

Remote · $80k – $100k / year · Full-time · Posted 4 months ago

This role is no longer accepting applications

Don't miss the next role like this

Create a free profile with your skills. We match you to new AI roles as they open, and companies hiring can contact you directly.

Create free profile, 2 minutes

Similar open roles you can apply to now:

See all open AI jobs

This role is no longer accepting applications

Don't miss the next role like this

Create a free profile with your skills. We match you to new AI roles as they open, and companies hiring can contact you directly.

Create free profile, 2 minutes

Similar open roles you can apply to now:

See all open AI jobs

About the role

Weld quality determines structural integrity in automotive chassis, aerospace components, pipelines, and heavy machinery. A weld with porosity, undercut, incomplete fusion, or excessive spatter can pass visual inspection from 2 meters away and fail catastrophically under load stress. Manual weld inspection is skilled labor — certified weld inspectors are expensive and in short supply — and doesn't scale to the throughput of modern automated welding operations. We build industrial visual inspection systems for 22 manufacturing clients. Weld inspection is one of the most technically demanding computer vision applications we've deployed: the defect features are small relative to the weld area, the appearance of a correct weld varies significantly by weld type and material, and the consequences of a false negative (passing a defective weld) are far more severe than a false positive (flagging a good weld for re-inspection). You'll build and maintain the weld seam inspection models: training defect detection models across the weld defect taxonomy (porosity, cracks, undercut, overlap, incomplete fusion), handling the appearance variation from different weld types (MIG, TIG, spot), and tuning the detection threshold strategy to meet the asymmetric cost requirement where false negatives must be minimized even at the cost of false positive rate.

Fully remote.

Responsibilities

Train weld defect detection models covering the standard weld defect taxonomy across weld types and materials
Build appearance variation handling for MIG, TIG, spot, and laser weld configurations
Implement detection threshold strategy minimizing false negatives under asymmetric cost requirements
Develop model adaptation pipeline for new weld configurations without full retraining
Create weld inspection report generation connecting defect detections to quality management systems
Monitor field false negative rates as the primary model quality metric

Requirements

3+ years in computer vision engineering with industrial inspection, defect detection, or quality control experience
Strong Python and PyTorch skills with object detection and segmentation model development
Understanding of weld metallurgy, defect taxonomy, and quality inspection standards (preferred)
Experience tuning detection thresholds for asymmetric precision-recall requirements
Ability to work with manufacturing engineers and certified weld inspectors on validation protocols