After tweaking an ML ensemble model for a few weeks it came time to deploy and in this case the project was utilizing an AWS pipeline. I had already dumped the trained model to an S3 bucket with joblib, so the goal now was to deploy an inference endpoint as an AWS Lambda function.

I previously deployed an ML inference on AWS via the Serverless framework and at the time I was using a scikit-learn model with a couple standard libraries...

Pushing an ML docker container to AWS Lambda made easy with Serverless.