Questions, Feedback, or just to discuss topics around kubeflow, cloud and cloudnative in Machine Learning projects? Let me know - wb@hypatos.ai.
Helpful, please share/like a LI post about this talk or just give a star to this github repo.
Description
A big part of Machine Learning projects is about engineering. We need to prepare data, build models, retrain them, not to mention about scaling the whole process and keeping it deterministic. Kubeflow helps us with it bringing best practices to manage and deploy such workflows. The second part of the engineering starts when the models are ready, and we need to bring them to production and operate them. Here we have ML model servers and Kfserving, that brings benefits of Knative and Istio to deploy, scale, and monitor our Machine Learning components.
The complete Kubeflow might be overkill for smaller teams in the beginning, thus we will also show how to start smaller and pick just necessary components from kubeflow community, such as Argo and tf serving.