Review on Swivl by Michael Sharp

The ability to train our model with different datasets from public domain, private domains or custom data in house without having access to expensive GPUs was one of my requirements when I started working for this solution provider. They have been extremely helpful throughout all stages starting at planning right through execution. There are no downsides as far we can see so far but they do charge quite high fees compared to other solutions like AWS etc. If you want an enterprise level SaaS platform which has full lifecycle support (including architecture) then look into it instead of going directly to cloud providers like amazon/azure. We were able to easily deploy models across multiple regions using their multi-region functionality builtin within their product while maintaining low latency between them by doing that. You also get very granular control over your infrastructure including VPC's.