Computer Vision Engineer

[What you'll do]
• Define 'perfection' in micro soldering: Translate 0.01mm-level laser joint quality into a visual language that AI can judge
• Fight against real-world noise: Build robust vision systems that work despite unstable lighting, reflective metal surfaces, and changing field environments
• Real-time inference on Edge: Optimize GPU/NPU-based real-time inspection algorithms to match production line speeds
• War against False Positives: Prove with data when the field says "this isn't a defect, it's a process characteristic" and reflect it in the model

[Who we're looking for]
• Someone who digs to the end asking "Why did this defect occur?" - Not just drawing bounding boxes, but someone with curiosity to understand optical phenomena and material properties
• A realist who knows perfect datasets don't exist - Someone who thinks "How far can we go with this data?" before saying "We don't have enough labels"
• Someone who gets excited seeing 10,000 photos of 1mm joints - The type who finds joy in discovering patterns in seemingly similar images
• Comfort Zone Exit: The courage to choose 95% stability in factory floors over 99% accuracy in research papers

[Nice to have]

Domain Knowledge
• Experience developing or operating AOI (Automated Optical Inspection) systems
• Experience with defect inspection in semiconductor/electronics manufacturing
• Understanding of thermal processes like laser processing, soldering, welding
• Basic knowledge of optics (lighting, lenses, camera parameters)

Technical Skills
• Computer Vision: OpenCV, PyTorch Vision, YOLO, MMDetection
• Defect Detection: Anomaly Detection, Segmentation, Instance Detection
• Edge AI: TensorRT, ONNX Runtime, OpenVINO
• Image Processing: Lighting correction, Morphology, Color space optimization

[Growth]

You'll grow from a vision engineer in papers to a craftsman who builds 'eyes' that actually work in the field.