Apache Airflow - asyncio in ETL task using TaskFlow API

Question

I'm trying to write ETL using airflow with asynchronous functionality.
I saw two examples in airflow official repo that have implemented ETL but didn't saw any async example.

Simple ETL https://github.com/apache/airflow/blob/master/airflow/example_dags/tutorial_etl_dag.py

TaskFlow API ETL: https://github.com/apache/airflow/blob/master/airflow/example_dags/tutorial_taskflow_api_etl.py

since TaskFlow API is much more better way to make the code readable I have tried to implement some async tasks using eventloop like in this question but without any success.

for example :

import json
import asyncio
from textwrap import dedent

from airflow import DAG

from airflow.operators.python import PythonOperator
from airflow.utils.dates import days_ago


default_args = {
  'owner': 'airflow',
}

with DAG(
  'tutorial_etl_dag',
  default_args=default_args,
  description='ETL DAG tutorial',
  schedule_interval=None,
  start_date=days_ago(2),
  tags=['example'],
) as dag:

def extract(**kwargs):
    ti = kwargs['ti']
    data_string = '{"1001": 301.27, "1002": 433.21, "1003": 502.22}'
    ti.xcom_push('order_data', data_string)

def transform(**kwargs):
    ti = kwargs['ti']
    extract_data_string = ti.xcom_pull(task_ids='extract', key='order_data')
    order_data = json.loads(extract_data_string)

    total_order_value = 0
    for value in order_data.values():
        total_order_value += value

    total_value = {"total_order_value": total_order_value}
    total_value_json_string = json.dumps(total_value)
    ti.xcom_push('total_order_value', total_value_json_string)

async def load(**kwargs):
    ti = kwargs['ti']
    total_value_string = ti.xcom_pull(task_ids='transform', key='total_order_value')
    total_order_value = json.loads(total_value_string)
    
    # upload total_order_value to cloud storage asynchronously
    # await upload_to_s3(total_order_value) 
    return total_order_value
    

def run_async():
   loop = asyncio.get_event_loop()
   result = loop.run_until_complete(load()) # kwargs ???? 
   return result

extract_task = PythonOperator(
    task_id='extract',
    python_callable=extract,
)

transform_task = PythonOperator(
    task_id='transform',
    python_callable=transform,
)

load_task = PythonOperator(
    task_id='load',
    python_callable=run_async,
)

extract_task >> transform_task >> load_task

in the example above the task "load_task" expects to receive kwargs argument (xcom data), how do I pass it from run_async method?

any ideas how to make it work?

Thanks!