I work for a startup that specializes in data matching (usually called reconciliation in banking / finance industry). Some core matching algorithms rely on comp

I work for a startup that specializes in data matching (usually called reconciliation in banking / finance industry). Some core matching algorithms rely on comparing sorted input using sliding windows; others rely on bucketing of some kind; anything to avoid O(m*n) when comparing two lists. The lists involved can easily run into the millions, depending on the domain.