Kind of
Math itself is omniversal
While the exact representations, rules, and applications may vary, the observational strategy of logic within mathematics apply not only to this universe but can apply to everything in all universes.
Geometry in other universes might not be Euclidean, but that doesn't rule it out as being non-mathematical. It would simply be based on different assumptions. Technically spacetime is non-Euclidean so this is actually the case for our own universe, however, Euclidian geometry is still extremely useful at the scales we typically operate.
While unlikely or perhaps impossible, you might have some universe where 1 + 1 = 2 and 2 + 2 = 4 but 2 + 1 + 1 = 5, but that just means certain numeric axioms no longer hold. Beings in such a universe could conceivably work with a mathematical system that assumes associativity of addition of integers even if their world doesn't actually work like that. After all, we use imaginary numbers even though there isn't really a physical analog for their meaning. (Though they are useful for certain physical properties like electromagnetism) (As a tangent, there actually are mathematical number spaces where addition is non-associative, but I don't know of any that have a practical application.)
The only way math can be useless in some universe is if it is so chaotic that there is no consistency of behavior and there are no universal laws. Such a universe could not support structure, let alone life.
But bringing it back to just our universe, it's extremely unlikely that, say, prime numbers will suddenly stop working in some remote corner of the universe. Geometry and algebra will still work elsewhere. The rules are going to be the same no matter where you are in the universe or even what universe you are in.
Notation is not universal
As far as communicating with aliens, the only barrier is representation. Humans have more or less agreed on the use of base 10 using arabic big-endian place-value numerals, symbols used for operators and common functions, using pi rather than tau, degrees, radians, and a couple of different notations for calculus (due to the Leibniz/Newton split). We've even decided on a binary encoding for all the symbols we ever would want to use. But all this was after thousands of years of development of humanity and most of the standardization has come in the last 500 years or so (with Unicode occupying the last 30 years or so).
Aliens almost certainly developed differently than humans, resulting in different representations of mathematical concepts. This wouldn't be a simple matter of different symbols to learn. For integers alone, there are hundreds of plausible notations from base, mixed-radixes, big and little endian representations, or even more complex representations of numbers like graphs. Real and complex numbers then multiply that further. Adding in operators, there are at least three plausible notations (prefix, postfix, and infix) if expressions are represented in a linear fashion. Expressions could also be represented in a tree structure. As you go through every mathematical concept, it's not unreasonable that you'll find some similar notation somewhere, but to have it match everywhere is extraordinarily unlikely.