Build a Simple ETL Pipeline (MLOps)

def extract(raw_data: list[dict]) -> list[dict]:
    extracted = []
    for record in raw_data:
        if record is not None and isinstance(record, dict):
            extracted.append(record)
    return extracted

def transform(records: list[dict], key_map: dict = None,
              numeric_fields: list[str] = None) -> list[dict]:
    transformed = []
    for record in records:
        new_record = {}
        for k, v in record.items():
            new_key = key_map.get(k, k) if key_map else k
            if numeric_fields and new_key in numeric_fields:
                try:
                    v = float(v)
                except (ValueError, TypeError):
                    v = 0.0
            if isinstance(v, str):
                v = v.strip().lower()
            new_record[new_key] = v
        transformed.append(new_record)
    return transformed

def load(records: list[dict]) -> dict:
    return {
        "data": records,
        "count": len(records),
        "columns": list(records[0].keys()) if records else []
    }

def etl_pipeline(raw_data: list[dict], key_map: dict = None,
                 numeric_fields: list[str] = None) -> dict:
    extracted = extract(raw_data)
    transformed = transform(extracted, key_map, numeric_fields)
    result = load(transformed)
    return result