OP's solution seems a little clunky to me. He does this thing where he loads the first day into a hashmap, and then he queries it as he loops over the second day. But why? That just overcomplicates your algorithm, and the memory used by O(2) days is roughly the same as O(1) days. It's also overly specialized to solving the very precise problem that's posed in the interview; what if tomorrow Big Boss wants you to aggregate over 3 days?
A cleaner solution would be to load both days 1 and 2 into the same hashmap. Then you can iterate the map and count whatever condition you want.
A cleaner solution would be to load both days 1 and 2 into the same hashmap. Then you can iterate the map and count whatever condition you want.