Oh the privilege of dealing with datasets measuring KB’s (apparently) and being able to use read.csv().
For the “real-world” data, one is forced to use fread. Sometimes passing to it custom-written awk statements that filter the required rows and columns from a csv.
You can also chunk with pandas in Python. But some unfortunates are forced to work with R (e.g., because the Python community didn’t get around creating an equivalent of Shiny).