Abstract: The new Petuum Platform is a composable platform for automatic and scalable MLOps. Today's widespread practice of ad-hoc integration between many fragmented ML tools leaves hard-to-fill gaps in end-to-end automation, scalability, and management of AI/ML applications. With the Petuum Platform, ML applications and infrastructure can be composed quickly and flexibly from standardized and reusable building blocks, thus transforming MLOps from ad-hoc craft production into a repeatable assembly-line process. This presentation and demo will show how your team can easily compose, manage, and monitor AI/ML infrastructure across multiple systems on a single pane of glass, seamlessly scale ML pipelines from local development to batch execution and online serving, and optimize end-to-end ML pipelines in an automatic and cost-efficient way. We will discuss new innovations in Composable, Automatic, and Scalable ML (CASL), developed in collaboration with CMU, UC Berkeley, and Stanford, and how they play a pivotal role in the Petuum Platform.
Bio: Aurick Qiao is the Chief Executive Officer at Petuum. Aurick received his Ph.D. from Carnegie Mellon University, where he researched distributed machine learning systems. His work on elastic scheduling for deep learning training recently won the Jay Lepreau Best Paper Award at OSDI 2021. Together with his experience at top technology companies such as Microsoft, Facebook, and Dropbox, Aurick is building products to support the next generation of AI/ML operations.