Have any of you implemented lightweight apps that use statistical/machine learning? What language(s)/framework(s) did you use and what would you suggest? Were there any optimization strategies that you employed and can share?

I'm limited by cloud costs so I would prefer such computations being done client side as much as possible thus I am trying to minimize time and memory complexity with desktops/notebooks. I am looking at all options in general but if you need specifics, my requirements eventually need to interface with Javascript even if it runs locally. I'm pretty much limited to the Python frameworks for work so am trying to identify and explore other options.

Obviously, this is not StackExchange, Quora, or Reddit but being that I am eager for information and this community has a few computer scientists and SDE's, I figured I'd try here as well.