Washington University in St. Louis

University of Science and Technology of China

Software Development Engineer at Amazon

Seattle, Washington, United States

Amazon Web Services (AWS)

Software Development Engineer

- Designed and implemented a machine learning model for interpretable time series classification based on dynamic routing and logistic regression with PyTorch and Python.- Tested and explored the interpretability, efficiency, accuracy of the time series classifier on 88 datasets from UCR time series dataset.- Enhanced the interpretability of time series classification task without losing classification accuracy.- Implemented an unsupervised learning model, Noise2Noise, for image restoration with TensorFlow and Python.- Built and pre-processed the MRI restoration dataset.-  Expanded Noise2Noise model as an application of compressed sensing in MRI and improved the quality and speed of MRI reconstruction.

Research Assistant

- Designed and implemented a deep neural network for real-time video object detection.- Refined the model to automatically schedule the key-frame based on optical flow, Res-Net, Faster RCNN, sequential NMS with MXNet, AWS and Python.- Evaluated and analyzed the model on ImageNet VID dataset.- Accelerated the detection speed and improved the detection accuracy, which all achieved state-of-the-art performance.

Wei Zhao