Jun
8
10:00pm
ML/AI on Kubernetes with Kubeflow
By IBM Developer
Keynote: Kubeflow overview and IBM
Speaker: Animesh Singh, Distinguished Engineer, IBM
Animesh will be sharing the work the team at IBM been doing with Kubeflow. Where we are and where we see the community headed.
Next Up:
Kubeflow pipelines with Tekton
Speakers: Tommy Chaoping Li, Software Developer, IBM, Yi Hong Wang, Software Developer and Performance Engineer, IBM
KubeFlow Pipelines (KFP) is a pipeline orchestration platform to simplify the pipeline building process and improve the Tekton Pipelines user experience with additional AI workflow features such as metric visualization, lineage tracking, and artifact metadata. KFP provides easy ways to create pipelines using the Python SDK and makes the pipelines simple, portable, and scalable, making it very popular among data scientists. KFP with Tekton is KFP running on the Tekton CI/CD pipeline that provides better Openshift integration and new concepts such as any sequencer, recursion, and pipeline loops on Tekton.
KFServing
Speakers: Paul van Eck, Software Engineer - Cognitive OpenTech, IBM, Andrew Butler, IBM
Machine learning model deployment can oftentimes be a challenging task for data scientists, and there are usually advanced requirements surrounding it such as data transformation and model explainability. In this talk, learn how KFServing, a Kubernetes-based model serving toolkit, can be used to tackle the majority of these use cases. From simple model deployments to deployments using explainers and transformers, here, we will demonstrate what capabilities KFServing has to offer.
MLX (The Machine Learning Exchange
Speaker: Christian Kadner, Software Developer, IBM
The Machine Learning Exchange (MLX), is a Data and AI Assets Catalog and Execution Engine. It allows data scientists to browse, download and upload ML/AI pipelines, pipeline components, models, datasets and notebooks. With automated sample pipeline code generation, the registered components, models, datasets and notebooks can be run with Kubeflow Pipelines on Tekton on a Kubernetes cluster. Featured technologies include Kubernetes, Kubeflow, KFServing, Datashim, Data Asset Exchange (DAX), Model Asset Exchange (MAX), MLSpec.
Kubeflow 1.3 on IKS/OpenShift
Speaker: Mofizur Rahman, Software Developer | Advocate, IBM
Demo of Kubeflow 1.3 on IKS and OpenShift
hosted by
IBM Developer
share