The operations referred to are likely
• raising to a power (the numerator of the exponent)
• taking a root (the denominator of the exponent)
• finding a reciprocal (because the exponent is negative)
They can be performed in any convenient order. It often works well to deal with small positive integers, so if one or more of these operations lets you proceed with a small positive integer for the remaining operations, that would be the one you'd perform first.
_____
A computer performing operations with a negative rational exponent may do so using logarithms. That is, the log of the base will be multiplied by the exponent, then the antilog found. The exponent itself will likely be treated as a floating point number, unless coding specifically indicates otherwise.