• Conduct research and experimentation on AI applications for project purposes in Janpan including Object Detection, Segmentation, Recognition and Video Classification.
• Build OCR and Document Layout Analysis pipelines (e.g., YOLO-based, OCR models) to extract annotations, dimensions, and labels from technical drawings and PDFs.
• Apply advanced Machine Learning techniques such as CNNs, Vision Transformers, and multi-label classification.
• Conduct model evaluation, benchmarking, hyperparameter tuning, and performance optimization.
• Prepare work reports and fulfill other requested tasks.

• Skills & Experience: Machine Learning & Computer Vision
• Hands-on experience in Object Detection, Face/Action Recognition, Image Segmentation, OCR, and Document Layout Analysis (e.g., PaddleOCR, LayoutLM, etc.)
• Solid understanding of Image Classification, Multi-label Classification, Semantic & Instance Segmentation
• Experience with model evaluation metrics such as mAP, IoU, F1-score, and optimization techniques
• Strong experience with PyTorch (preferred); familiarity with OpenCV, Scikit-learn, Pandas, Albumentations
• Experience working with datasets and labeling tools such as Label Studio, CVAT, or similar

Nice to Have

• Experience with model serving and deployment tools: ONNX Runtime, Triton Inference Server, TorchServe
• Knowledge of production ML systems: low-latency inference optimization, retraining pipelines, model drift monitoring
• Experience with C++ is a plus