Date: September 29, 2022 03:00 PM Pacific Time (US and Canada)
- Welcome, Announcements & Housekeeping
- Talk #1: Intro to Kubeflow
Co-organizer, Jimmy Guerrero will give a 10-min, broad overview of the open source Kubeflow MLOps platform. We’ll cover architecture, components, distributions and installation options.
- Talk #2: Blending AI Techniques for Demand and Supply Planning Use
Manufacturers and retailers constantly face a highly complex challenge: how to predict demand followed very closely by how to allocate limited supply to often volatile, demand. Traditionally, the tools available to practitioners (or planners) have relied heavily on mutually exclusive techniques (e.g. heuristics OR linear programs OR machine learning, etc.) to solve these problems. With the proliferation of easy access to compute and storage, the “mono-algorithm” approach is starting to change – enabling a “compositional” approach to algorithm construction. Waleed will be discussing these “composite” algorithmic structures, showing some examples and the demands they place on MLOps-like architectures.
Waleed Ayoub was recently the CTO at Rubikloud, a machine learning enterprise software company that was acquired by Kinaxis in 2020, where he spent 2 years as an SVP of product development.
- Talk #3: A Deep-Dive Into Declarative AutoML on Kubernetes
When dealing with ever-increasing requirements and growing numbers of models, data scientists may often turn to AutoML as a solution for minimizing technical debt and optimizing the time and effort required to train and deploy models to production. In this session, Software Engineer Tomer Sagi will introduce the theory behind AutoML and the various different techniques for optimizing your model’s algorithm and hyper-parameters. Furthermore, he will discuss how AutoML can be applied to other areas of the machine learning process in what makes an end-to-end ML system. He will also explore how ML systems are implemented in the context of Kubernetes with platforms like Kubeflow, as well as Modela, a new declarative ML system based on Kubernetes Custom Resources.
Tomer Sagi is the founder of Metaprov and was a senior developer most recently at HPE where he designed and developed a file system services as part of HPE storage products.