{
  // Workflow metadata
  "name": "data_pipeline",

  "description": "Multi-stage data processing pipeline for multiple datasets",

  // Job definitions
  "jobs": [
    {
      "name": "download_{dataset}",
      "command": "wget https://example.com/data/{dataset}.csv -O /data/raw/{dataset}.csv",
      "resource_requirements": "download",
      "output_files": [
        "raw_{dataset}"
      ],
      "parameters": {
        "dataset": "['sales','inventory','customers','products']"
      }
    },
    {
      "name": "preprocess_{dataset}",
      "command": "python scripts/preprocess.py --input=/data/raw/{dataset}.csv --output=/data/clean/{dataset}.csv",
      "resource_requirements": "cpu_intensive",
      "depends_on": [
        "download_{dataset}"
      ],
      "input_files": [
        "raw_{dataset}"
      ],
      "output_files": [
        "clean_{dataset}"
      ],
      "parameters": {
        "dataset": "['sales','inventory','customers','products']"
      }
    },
    {
      "name": "analyze_{dataset}_{analysis_type}",
      "command": "python scripts/analyze.py \\\n  --input=/data/clean/{dataset}.csv \\\n  --analysis={analysis_type} \\\n  --output=/results/{dataset}_{analysis_type}_results.json\n",
      "resource_requirements": "cpu_intensive",
      "depends_on": [
        "preprocess_{dataset}"
      ],
      "input_files": [
        "clean_{dataset}"
      ],
      "output_files": [
        "results_{dataset}_{analysis_type}"
      ],
      "parameters": {
        "dataset": "['sales','inventory','customers','products']",
        "analysis_type": "['summary','trends','anomalies']"
      }
    },
    {
      "name": "report_{dataset}",
      "command": "python scripts/generate_report.py \\\n  --dataset={dataset} \\\n  --input-dir=/results \\\n  --output=/reports/{dataset}_report.html\n",
      "resource_requirements": "minimal",
      "depends_on": [
        "analyze_{dataset}_{analysis_type}"
      ],
      "input_files": [
        "results_{dataset}_{analysis_type}"
      ],
      "parameters": {
        "dataset": "['sales','inventory','customers','products']",
        "analysis_type": "['summary','trends','anomalies']"
      }
    },
    {
      "name": "executive_summary",
      "command": "python scripts/executive_summary.py --input-dir=/reports --output=/reports/executive_summary.pdf",
      "resource_requirements": "minimal",
      "depends_on": [
        "report_{dataset}"
      ],
      "parameters": {
        "dataset": "['sales','inventory','customers','products']"
      }
    }
  ],
  
  // File definitions
  "files": [
    {
      "name": "raw_{dataset}",
      "path": "/data/raw/{dataset}.csv",
      "parameters": {
        "dataset": "['sales','inventory','customers','products']"
      }
    },
    {
      "name": "clean_{dataset}",
      "path": "/data/clean/{dataset}.csv",
      "parameters": {
        "dataset": "['sales','inventory','customers','products']"
      }
    },
    {
      "name": "results_{dataset}_{analysis_type}",
      "path": "/results/{dataset}_{analysis_type}_results.json",
      "parameters": {
        "dataset": "['sales','inventory','customers','products']",
        "analysis_type": "['summary','trends','anomalies']"
      }
    }
  ],
  
  // Resource requirements
  "resource_requirements": [
    {
      "name": "minimal",
      "num_cpus": 1,
      "num_gpus": 0,
      "num_nodes": 1,
      "memory": "2g",
      "runtime": "PT10M"
    },
    {
      "name": "download",
      "num_cpus": 1,
      "num_gpus": 0,
      "num_nodes": 1,
      "memory": "1g",
      "runtime": "PT15M"
    },
    {
      "name": "cpu_intensive",
      "num_cpus": 8,
      "num_gpus": 0,
      "num_nodes": 1,
      "memory": "16g",
      "runtime": "PT1H"
    }
  ],

}