Hi, @Joyce_Tam and than you for the response, @sotiri!
I did try to do it the way @sotiri suggested but was unsuccessful. I tried to load participant-specific .csv files at the start and then have only the resources in those files load on the fly, but that did not work for me. It was a few months ago, so I cannot remember exactly what my errors were, but I ended up just pre-generating a random subset of my stimuli that participants and just randomized that pre-generated subset (so all participants saw the same random subset but in a random order) for each person so to keep the download time manageable.