{
  "context": "This criteria evaluates how well the engineer uses Deep Lake's storage abstraction capabilities to build a dataset manager that works seamlessly across different storage backends (local filesystem, S3, GCS, Azure). The focus is on proper usage of Deep Lake's unified storage API for dataset lifecycle operations.",
  "type": "weighted_checklist",
  "checklist": [
    {
      "name": "Dataset creation",
      "description": "Uses deeplake.create() to create datasets at specified paths, accepting both local and cloud storage URLs",
      "max_score": 20
    },
    {
      "name": "Existence checking",
      "description": "Uses deeplake.exists() to check if a dataset exists at a given path before creating or operating on it",
      "max_score": 15
    },
    {
      "name": "Schema definition",
      "description": "Uses dataset.add_column() with appropriate types (e.g., types.Text(), types.Int32()) to define the dataset schema as specified",
      "max_score": 15
    },
    {
      "name": "Data insertion",
      "description": "Uses dataset.append() or dataset.extend() to add sample data rows to the dataset",
      "max_score": 15
    },
    {
      "name": "Dataset copying",
      "description": "Uses deeplake.copy() to copy datasets between storage locations (source to destination)",
      "max_score": 20
    },
    {
      "name": "Dataset deletion",
      "description": "Uses deeplake.delete() to remove datasets from storage when cleanup is needed",
      "max_score": 15
    }
  ]
}