You have more experience in this area, but REGEXP_CONTAINS() seems faster than LIKE for bigger datasets IMO.

Although in my work I tend to use REGEXP_CONTAINS() as an efficient multifilter for different inputs, which speeds things up too.