Skip to content

OpenShift AI Platform Analysis

Architecture snapshot: 2026-05-20 (2026-05-20)

Generated by architecture-analyzer. All data produced by deterministic static analysis.

Platform Summary

Metric Count
Components 56
CRDs 56
Services 81
Secrets 45
Cluster Roles 61
Cross-Component Dependencies 24
Webhooks 121

Component Dependency Graph

%%{init: {'theme': 'base', 'flowchart': {'nodeSpacing': 30, 'rankSpacing': 50}}}%%
graph LR
    classDef internal fill:#3498db,stroke:#2980b9,color:#fff
    classDef external fill:#95a5a6,stroke:#7f8c8d,color:#fff

    ai_gateway_payload_processing["ai-gateway-payload-processing\n2 CRDs"]:::internal
    data_science_pipelines_operator["data-science-pipelines-operator\n4 CRDs"]:::internal
    gateway_api_inference_extension["gateway-api-inference-extension\n4 CRDs"]:::internal
    kserve["kserve\n26 CRDs"]:::internal
    kubeflow["kubeflow"]:::internal
    llm_d_inference_scheduler["llm-d-inference-scheduler"]:::internal
    model_registry["model-registry"]:::internal
    modelmesh_serving["modelmesh-serving\n4 CRDs"]:::internal
    models_as_a_service["models-as-a-service"]:::internal
    odh_cli["odh-cli"]:::internal
    opendatahub_operator["opendatahub-operator"]:::external
    spark_operator["spark-operator\n3 CRDs"]:::internal
    workload_variant_autoscaler["workload-variant-autoscaler\n1 CRDs"]:::internal

    ai_gateway_payload_processing ==>|"watches 3 CRDs"| gateway_api_inference_extension
    kserve ==>|"watches 3 CRDs"| gateway_api_inference_extension
    kubeflow -.->|"go-module"| data_science_pipelines_operator
    llm_d_inference_scheduler ==>|"watches 3 CRDs"| gateway_api_inference_extension
    model_registry ==>|"watches InferenceService"| kserve
    modelmesh_serving ==>|"watches 4 CRDs"| kserve
    models_as_a_service -.->|"go-module"| kserve
    models_as_a_service ==>|"watches ExternalModel"| ai_gateway_payload_processing
    odh_cli -.->|"go-module"| opendatahub_operator
    workload_variant_autoscaler ==>|"watches InferencePool, InferenceObjective"| gateway_api_inference_extension
    kserve -->|"webhook-ref"| modelmesh_serving
    modelmesh_serving -->|"webhook-ref"| workload_variant_autoscaler
    spark_operator -->|"webhook-ref"| workload_variant_autoscaler

Components Analyzed

Component CRDs
MLServer 0
NeMo-Guardrails 0
ai-gateway-payload-processing 2
ai4rag 0
argo-workflows 0
batch-gateway 0
caikit-nlp 0
caikit-tgis-serving 0
codeflare-sdk 0
data-science-pipelines 3
data-science-pipelines-operator 4
distributed-workloads 0
eval-hub 0
feast 0
fms-guardrails-orchestrator 0
fms-hf-tuning 0
gateway-api-inference-extension 4
guardrails-regex-detector 0
kserve 26
kube-auth-proxy 0
kube-rbac-proxy 0
kubeflow 0
kuberay 0
kueue 2
llama-stack 0
llama-stack-k8s-operator 2
llm-d 0
llm-d-inference-scheduler 0
llm-d-kv-cache 0
llm-d-routing-sidecar 0
lm-evaluation-harness 0
ml-metadata 0
mlflow-operator 2
model-metadata-collection 0
model-registry 0
modelmesh-runtime-adapter 0
modelmesh-serving 4
models-as-a-service 0
notebooks 0
notebooks-downstream 0
odh-cli 0
odh-deployer 0
openvino_model_server 0
pipelines-components 0
rest-proxy 0
rhds-llama-stack-distribution 0
spark-operator 3
text-generation-inference 0
trainer 3
trustyai-explainability 0
vllm-cpu 0
vllm-gaudi 0
vllm-orchestrator-gateway 0
vllm-rocm 0
vllm-spyre 0
workload-variant-autoscaler 1