Ong Kuan Lin

I’m an AI R&D Engineer focused on turning complex inspection datasets into reliable, production-ready intelligence. At Vitrox Corporation Berhad, I build preprocessing and feature engineering pipelines, run statistical analysis to spot defect patterns, and translate findings into process optimization.

I lead AI-based Automated Optical Inspection (AOI) software development for SMT components, designing defect inspection algorithms for bad joints, component offsets, missing parts, and solder defects. I develop and train deep learning models for object detection, image segmentation, and similarity learning—using tools like PyTorch and TensorFlow—to improve robustness across diverse datasets.

To drive measurable impact, I also design data-driven dashboards and analytical visualizations with VONE and PostgreSQL, enabling real-time defect trend analysis and contributing to over 40% improvement in First Pass Yield (FPY). I’ve supported 100+ customers globally with technical support, training, model deployment assistance, and database configuration, and I’ve previously worked as a Data Scientist building LLM-based PO extraction systems with over 90% field extraction accuracy and ~70% manual processing reduction.