AWS avoid loading to memory when putting

I’m recording videos from a raspberry pi and want to put those on an S3 bucket using AWSS3.jl. The thing is, these videos can be large approaching the RAM of the RPI (in my case 2GB). So I’s like to avoid loading them to memory, as a vector of bytes, which is what AWSS3.s3_put expects.

Is there a way to either:

  1. pipe the output of the camera (or anything really) directly into the bucket?
  2. put a saved file in the bucket without reading it whole into memory?