Implementation:SeldonIO Seldon core Seldon Pipeline CRD

Field	Value
Implementation Name	Seldon Pipeline CRD
Type	Pattern Doc
Overview	Concrete pattern for declaring Seldon Core 2 Pipeline resources as Kubernetes YAML manifests.
Related Principle	SeldonIO_Seldon_core_Pipeline_Topology_Definition
Source	samples/pipelines/tfsimples.yaml:L1-16
Domains	MLOps, Data_Flow
External Dependencies	Kubernetes API (mlops.seldon.io/v1alpha1), Kafka
Knowledge Sources	Repo (https://github.com/SeldonIO/seldon-core), Doc (https://docs.seldon.io/projects/seldon-core/en/v2/)
Last Updated	2026-02-13 00:00 GMT

Description

This implementation provides the concrete YAML pattern for defining Seldon Core 2 Pipeline custom resources. A Pipeline CRD declares a directed acyclic graph of model steps with explicit data flow edges and optional tensor name remapping. The Kubernetes API server validates the CRD schema, and the Seldon scheduler reconciles the desired state.

Code Reference

Pipeline CRD YAML (tfsimples)

apiVersion: mlops.seldon.io/v1alpha1
kind: Pipeline
metadata:
  name: tfsimples
spec:
  steps:
    - name: tfsimple1
    - name: tfsimple2
      inputs:
      - tfsimple1
      tensorMap:
        tfsimple1.outputs.OUTPUT0: INPUT0
        tfsimple1.outputs.OUTPUT1: INPUT1
  output:
    steps:
    - tfsimple2

Source: samples/pipelines/tfsimples.yaml:L1-16

Fan-in Pipeline YAML (join)

apiVersion: mlops.seldon.io/v1alpha1
kind: Pipeline
metadata:
  name: join
spec:
  steps:
    - name: tfsimple1
    - name: tfsimple2
    - name: tfsimple3
      inputs:
      - tfsimple1.outputs.OUTPUT0
      - tfsimple2.outputs.OUTPUT1
      tensorMap:
        tfsimple1.outputs.OUTPUT0: INPUT0
        tfsimple2.outputs.OUTPUT1: INPUT1
  output:
    steps:
    - tfsimple3

Source: samples/pipelines/tfsimples-join.yaml:L1-18

Key Parameters

Parameter	Description	Example
spec.steps[].name	Name of a deployed model to use as a pipeline step	tfsimple1
spec.steps[].inputs	List of upstream step references providing input data	[tfsimple1]
spec.steps[].tensorMap	Tensor name remapping from upstream outputs to this step's inputs	tfsimple1.outputs.OUTPUT0: INPUT0
spec.output.steps	List of step names whose outputs form the pipeline response	[tfsimple2]

I/O Contract

Inputs

• Deployed model names: Each step references a model that must already be loaded and available via the Seldon scheduler.
• Desired DAG topology: The logical flow of data between models, expressed as step input references.

Outputs

• Pipeline CRD YAML manifest: A Kubernetes-native resource definition that can be applied via seldon pipeline load or kubectl apply.

Usage Examples

Simple Two-Step Chain

In this pattern, tfsimple1 receives the pipeline input, produces OUTPUT0 and OUTPUT1, and tfsimple2 receives those outputs remapped to INPUT0 and INPUT1:

spec:
  steps:
    - name: tfsimple1
    - name: tfsimple2
      inputs:
      - tfsimple1
      tensorMap:
        tfsimple1.outputs.OUTPUT0: INPUT0
        tfsimple1.outputs.OUTPUT1: INPUT1
  output:
    steps:
    - tfsimple2

Fan-in from Two Parallel Steps

In this pattern, tfsimple1 and tfsimple2 both receive the pipeline input independently. tfsimple3 then joins their outputs:

spec:
  steps:
    - name: tfsimple1
    - name: tfsimple2
    - name: tfsimple3
      inputs:
      - tfsimple1.outputs.OUTPUT0
      - tfsimple2.outputs.OUTPUT1
      tensorMap:
        tfsimple1.outputs.OUTPUT0: INPUT0
        tfsimple2.outputs.OUTPUT1: INPUT1
  output:
    steps:
    - tfsimple3

Loading the Pipeline

seldon pipeline load -f ./pipelines/tfsimples.yaml

Related Pages

• SeldonIO_Seldon_core_Pipeline_Topology_Definition - implements - Principle describing DAG-based inference pipeline topology.
• SeldonIO_Seldon_core_Seldon_Model_Load_For_Pipeline - prerequisite - Models must be loaded before the pipeline CRD can be applied.
• SeldonIO_Seldon_core_Seldon_Pipeline_Advanced_Routing - extends - Advanced conditional routing patterns layered on the base CRD structure.
• SeldonIO_Seldon_core_Seldon_Pipeline_Load - next step - CLI tool for submitting the pipeline CRD to the scheduler.
• Environment:SeldonIO_Seldon_core_Kubernetes_Cluster_Environment
• Environment:SeldonIO_Seldon_core_Go_Build_Toolchain_Environment
• Environment:SeldonIO_Seldon_core_Kafka_Messaging_Environment