Building a cost optimization checklist
You should constantly pay attention to cost, even in the early stages of your machine learning project. Even if you're not paying the AWS bill, someone is, and I'm sure you'll quite quickly find out who that person is if you spend too much.
Regularly going through the following checklist will help you spend as little as possible, get the most machine learning-happy bang for your buck, and hopefully keep the finance team off your back!
Optimizing costs for data preparation
With so much focus on optimizing training and deployment, it's easy to overlook data preparation. Yet, this critical piece of the machine learning workflow can incur very significant costs.
Tip #1
Resist the urge to build ad hoc ETL tools running on instance-based services.
Obviously, your workflows will require data to be processed in a custom fashion, such as applying domain-specific feature engineering. Working with a managed service such...
