This message was deleted.

Hey @Matthieu Vanhoutte If your team is using kubernetes or AWS EKS(elastic kubernetes service), build docker image from bento bundle is probably the best way forward. Sagemaker potentially is a good fit for you, you can change the instance size and instance counts. You can deploy to sagemaker with bentoml CLI. You can find more info with

bentoml sagemaker deploy --help

AWS lambda is more suit for smaller model with lightweight workload. From your description, I don't think it is the best services for it at the moment.

Hi @Bo thanks for your feedback! I am effectively alone on my small team. It seems so that Sagemaker would be a good option. However, EC2 seems to cost less than sagemaker and bentoML has a tuto on deploying onto EC2 (https://docs.bentoml.org/en/latest/deployment/aws_ec2.html). Would it be feasible for a newbie to deploy onto AWS EC2 ?

@Matthieu Vanhoutte Yes. ec2 would be a great option as well. YOu can find more info with

bentoml ec2 deploy --help

Thanks you @Bo