Rscript -e 'df <- read.csv("data.csv"); print(quantile(df$salary))'

quantile() returns the 0, 25, 50, 75, and 100 percent percentiles of a vector, the classic five-number summary. Percentiles describe the distribution's shape and spread far better than a single average. Add probs=c(0.05, 0.95) to request custom quantiles instead. This is the R counterpart of the datamash perc: operations.