[Leadership] - As both AI project owner and core developer, managing the entire lifecycle from idea to delivery. - Key driver of cross-functional collaboration, aligning with user, algorithm, and system teams. [Deliveries] - Developed a python package for IP power consumption prediction by Linear-Tree, improved prediction from MAE 15% to 5% while providing model interpretability. - Developed an AI-based timing prediction web-app to increase timing granularity over 50x while keeping acceptable accuracy. - Build a video subtitle&voice translation tool using speech-to-text/LLM/text-to-speech models, saving 70% manual time.

- Developed Bread Detection App on POS machines for smart checkout with a few-shot detection model detecting new goods, resulting in over 85% 3-shot accuracy. - Building Customer Journey Analysis Tool from 0 to 1 - Survey and implement 3DCNN-based models to analyze customer activities. - Incorporate human-human interactions with GCNs into activity recognition models. - Training Multi-Object Tracking Models to track items for smart packing checks. - Research on model deployment to devices. Experienced Onnx, TFLite, Openvino, SNPE, TensorRT. - Developed labeling tools for merging multi-camera view customer tracklets.