Whenever you're ready, walk me through how you'd approach this.
How do you handle audit sampling when dealing with large, high-volume datasets?