Automated Machine Learning Pipeline with Mesos

Course details

Mesos, with its semi-centralized infrastructure, sustains the skeleton of Silicon Valleys Netflix (Fezo), Airbnb (Airflow), Heroku, and Apple to name a few, and has established itself as a staple in any automated machine learning pipeline and distributed heterogeneous data pruning.

In this course, we will learn the foundation of Mesos within the automated pipeline on fault-tolerant cluster semaphores. We will set up a virtual cluster running Marathon and Zookeeper and a concurrent Docker application. We will establish a master-slave infrastructure, experience real-time debugging, and learn how to automate cluster arbitration via Soliton automata. We will then see an iterative queue manager for indexed tasks dispatched concurrently inside a poset topology.

About The Author

Karl Whitford has been involved in the tech industry for 10 years as a software engineer. He has a background in statistical machine learning, deep learning, and A.I. from Columbia University. He also has knowledge of computational physics/mathematics from DePaul University and UT Austin. He is a professional in the fields of game A.I, compression, machine learning, and distributed cluster computing. Karl is an open source contributor to SMACK, Pancake Stack (PipelineI/O), and Pregel-Mesos, among others. He has previous work experience with Microsoft, Coca Cola, and Unilever to name a few; he is also an indie game developer and founder of Esquirel (Black-Squirrel) Studios in San Francisco, California. He was also handpicked by UploadVR as "one to watch" and featured at Mountain Views 2016 VR Showcase.