Beetlejuice. A script that takes a file and transcodes is using temporary cloud compute.

Current structure:

• Run mediainfo on input, output to XML
• Spin up a compute node
• Upload input file
• Run Handbrake-CLI using a mix of parameters and mediainfo based off the ptp transcoding rules
• Upload resulting transcode somewhere
• Delete compute node

Inputs:

• Filename
• API key
• Encoding options (how many?)
• Upload location & credentials (keys only?)