In the lesson prerequisites, access to an Nvidia GPU seems required or access to Google Colab. Maybe there are also restrictions on Cuda version supported or compute capability?
The image processing example and relation to astronomy is nice. As Python seems to be the main prerequisite, it may be helpful to assume background provided by the Carpentries Python lessons. As such, a little more explanation on computer architecture and algorithms may be helpful - in particular describing algorithmic patterns that will work well on a GPU
May also want to explain energy efficiency of GPUs not just speedup
A large example has been added to episode 2, so if anyone is willing to read and review it, that would be appreciated.