I'm working on Belle II and ATLAS where for ATLAS i'm involved in getting columnar data analysis working with DAOD PHYSLITE. For this i implemented a schema class for coffea.nanoevents which i'd like to get awkward_dask ready during or soon after the workshop. For Belle II i'm working with students on ML projects where they often come to me for help when they want to preprocess "awkward" data, e.g. for models trained on 4-momentum vectors. I'm also interested in pyhf which seems to get more and more used at Belle II as well.