Scalable Low-Latency Model Prediction for VideoAI/ML и визуализация данных
Ben studied finance at Emory University, and is currently a data scientist at Mux. While at Mux, he has worked on computer vision and deep learning projects for the past two years. His past experiences include operations, analytics, and consulting at Google and Deloitte.
This talk will discuss process through which Mux setup our video prediction and training pipeline. We will look at a variety of technologies Mux evaluated for serving our models (including TF serving, Kubeflow, and Clipper) that resulted in scalable low-latency predictions, and how to overcome an array of challenges you might encounter when setting up your own data pipelines, specifically when it comes to processing data related to images and video.