Here you will find some PoCs that I've developed using AWS.

First of all, build your own Airflow image.
This command will add your AWS credentials to your Airflow container and automatically it will configure the default AWS connection.

docker-compose build \
  --build-arg ACCESS_KEY=$(cat ~/.aws/credentials | grep key_id | sed 's/.*key_id.*= //') \
  --build-arg SECRET_KEY=$(cat ~/.aws/credentials | grep secret | sed 's/.*secret.*= //')

Now, let's up a container with our image.

# Airflow will be deployed at "http://localhost:8080"
docker-compose up -d

When you finish using, just put down the container.

docker-compose down

Dags
At docker-compose.yml there are volumes pointing to every PoC DAG folder. If you have a new PoC and want to map your new DAG folder you have to create another volume mapping at docker-compose.yml.

Default Variables
At airflow/config/variables.json you will find all variables that are going to be imported into Airflow at startup. If you need to add another one, just add it to variables.json and rebuild the image. Be careful with secrets, I don't recommend to store these kind of values in this json file!