# Pipelines (/docs/guides/developer-workspace/pipelines)



Pipelines lets you define, run, and track multi-step machine learning and data processing
workflows. You write a pipeline in Python using [the Kubeflow Pipelines SDK](https://www.kubeflow.org/docs/components/pipelines/user-guides/core-functions/compile-a-pipeline/), compile it to
a YAML file, and upload that file to BullSequana AI. The platform executes
each step in the correct order, tracks every run, and stores the outputs for
later comparison.

<Cards>
  <Card title="Concepts" href="/docs/guides/pipelines/concepts" />

  <Card title="Upload a pipeline" href="/docs/guides/pipelines/upload-a-pipeline" />

  <Card title="Experiments" href="/docs/guides/pipelines/experiments" />

  <Card title="Runs" href="/docs/guides/pipelines/runs" />

  <Card title="Recurring runs" href="/docs/guides/pipelines/recurring-runs" />

  <Card title="Artifacts" href="/docs/guides/pipelines/artifacts" />

  <Card title="Executions" href="/docs/guides/pipelines/executions" />
</Cards>