Are you proposing brute forcing it? Depending on requirements that might be fine, but I doubt FB would like that kind of solution. If they did, they'd have a different kind of question.
abc , adc
a - 1
b - 1
c - 1
after second word loop
b - 1, d
Regardless, I was also thinking Levenshtein is sub-optimal, and that you can probably solve it in O(n+m).
Even if you go for Levenshtein first, you should identify a simple modification to stop calculating the matrix early if the distance is necessarily greater than 1. Get a bit fancier and you can just 'go down the diagonal' and greatly optimize the algorithm.