We already have a function _searchindex to search for subsequences of byte arrays. It would be nice to expose this by adding corresponding methods to findfirst and findnext.
Although the current _seardhindex implementation only accepts Vector arguments, it looks like the implementation doesn't depend on the storage format, so the function signature could be easily widened to AbstractVector{<:Union{Int8,UInt8}} arguments. (You'll need to replace the sizeof calls with length, and maybe use firstindex to support non 1-based indexing.)
(See also the discussion on discourse.)