Awkward Array: Manipulating JSON like Data with NumPy like Idioms |SciPy 2020| Jim Pivarski 

Awesome work and a great presentation!

@6:20 "at a time, Fortran didn't even have a for loop". The for loop in Fortran is called "do loop". It was always there.

Thanks Jim! As always, very nice work :)

Wow. So impressive. I’m installing awkward now.

since you have kind of a tree structure... can you also generalise this to a graph structure? Graphs with attached data I think have similar problems... and another similarity is with sparse matrices or should you call it sparse data? Just as a random input