{
  "id": "workflow:rag-pipeline-evaluation",
  "_kind": "Workflow",
  "_file": "workflows/workflows/workflows-ai-agents-deep.yaml",
  "_cluster": "workflows",
  "attributes": {
    "displayName": "RAG Pipeline Evaluation",
    "workflowKind": "data",
    "triggerType": "event-driven",
    "typicalCadence": "per-release",
    "complexity": "cross-team",
    "description": "Evaluates retrieval-augmented generation pipeline quality end-to-end —\nmeasuring retrieval precision@k and recall@k against ground-truth relevance\njudgments, scoring generation faithfulness and hallucination rates via LLM\njudges, benchmarking chunking and embedding strategies, profiling latency\nacross retrieval and generation stages, and running regression suites before\npromoting pipeline changes. Excludes RAG architecture design and embedding\nmodel training.\n"
  },
  "outgoingEdges": [
    {
      "from": "workflow:rag-pipeline-evaluation",
      "to": "role:ml-engineer",
      "kind": "involves_role",
      "attributes": {}
    },
    {
      "from": "workflow:rag-pipeline-evaluation",
      "to": "role:ai-champion",
      "kind": "involves_role",
      "attributes": {}
    },
    {
      "from": "workflow:rag-pipeline-evaluation",
      "to": "role:data-scientist",
      "kind": "involves_role",
      "attributes": {}
    },
    {
      "from": "workflow:rag-pipeline-evaluation",
      "to": "skill-area:retrieval-augmented-generation",
      "kind": "requires_skill_area",
      "attributes": {}
    },
    {
      "from": "workflow:rag-pipeline-evaluation",
      "to": "skill-area:eval-driven-development",
      "kind": "requires_skill_area",
      "attributes": {}
    },
    {
      "from": "workflow:rag-pipeline-evaluation",
      "to": "skill-area:prompt-engineering",
      "kind": "requires_skill_area",
      "attributes": {}
    },
    {
      "from": "workflow:rag-pipeline-evaluation",
      "to": "domain:ml-ops",
      "kind": "applies_to_domain",
      "attributes": {}
    },
    {
      "from": "workflow:rag-pipeline-evaluation",
      "to": "domain:data-science",
      "kind": "applies_to_domain",
      "attributes": {}
    },
    {
      "from": "workflow:rag-pipeline-evaluation",
      "to": "responsibility:ai-tooling-evaluation",
      "kind": "triggers_responsibility",
      "attributes": {}
    },
    {
      "from": "workflow:rag-pipeline-evaluation",
      "to": "responsibility:data-quality-monitoring",
      "kind": "triggers_responsibility",
      "attributes": {}
    },
    {
      "from": "workflow:rag-pipeline-evaluation",
      "to": "org-unit:ai-enablement",
      "kind": "performed_by_org_unit",
      "attributes": {}
    },
    {
      "from": "workflow:rag-pipeline-evaluation",
      "to": "org-unit:ml-team",
      "kind": "performed_by_org_unit",
      "attributes": {}
    },
    {
      "from": "workflow:rag-pipeline-evaluation",
      "to": "org-unit:research-engineering",
      "kind": "performed_by_org_unit",
      "attributes": {}
    }
  ],
  "incomingEdges": [
    {
      "from": "stack-profile:agentic-rag",
      "to": "workflow:rag-pipeline-evaluation",
      "kind": "follows_workflow"
    }
  ]
}