One of the most popular ways to train and serve models is to do AI/ML on Kubernetes. However, data scientists typically don’t do double duty as platform managers.
Standing up a Kubernetes environment can be a challenge for data scientists but most data scientists know Python. We will demonstrate how with Pulumi you can deploy a machine learning platform on GKE and leverage the power of GPUs and TPUs.
If you want to code along, you’ll need the following:
-- Create a Pulumi account: https://pulumip.us/Free-Sign-Up -- A Google Cloud account (free tier is okay).
Pulumi’s mission is to democratize the cloud for every engineer.
Its open-source Infrastructure as Code tool enables engineers to write infrastructure code in any programming language. Build infrastructure intuitively on any cloud with TypeScript, JavaScriptPython, Go, C#, Java, and YAML.