index.rst

Getting Started

Get up and running with Kubeflow SDK in minutes.

What is Kubeflow SDK?

Kubeflow SDK is a Python library that makes it easy to:

• Train ML models on your laptop or Kubernetes cluster
• Scale training from 1 GPU to hundreds without changing your code
• Tune hyperparameters to find the best model configuration

Think of it as "PyTorch Lightning for Kubernetes" - you write normal Python training code, and Kubeflow handles the infrastructure.

Who is this for?

• Data Scientists who want to scale their training without learning Kubernetes
• ML Engineers who need reliable distributed training
• Teams who want reproducible ML workflows

Quick Example

Here's how simple it is to train a model:

from kubeflow.trainer import TrainerClient
from kubeflow.trainer.types import CustomTrainer

def train():
    """Your normal training code - nothing special needed."""
    import torch

    model = torch.nn.Linear(10, 1)
    optimizer = torch.optim.SGD(model.parameters(), lr=0.01)

    for epoch in range(100):
        loss = model(torch.randn(32, 10)).sum()
        loss.backward()
        optimizer.step()
        print(f"Epoch {epoch}, Loss: {loss.item():.4f}")

# Submit to Kubernetes - that's it!
client = TrainerClient()
job_name = client.train(trainer=CustomTrainer(func=train))

# Wait and get logs
client.wait_for_job_status(job_name)

Next Steps

.. grid:: 3
   :gutter: 3

   .. grid-item-card:: Installation
      :link: installation
      :link-type: doc

      Install the SDK and verify it works.

   .. grid-item-card:: Quickstart
      :link: quickstart
      :link-type: doc

      Train your first model step-by-step.

   .. grid-item-card:: Local Development
      :link: local-development
      :link-type: doc

      Run training jobs locally using different SDK backends.