The scope of such a review would necessarily be limited to the correctness of the implementation rather than the strength of the algorithm.
However, as well as the usual Code Review items (such as readability, program structure, speed/memory performance, error handling and the rest), there are opportunities to consider some additional aspects of the implementation:
- weakness to timing attacks
- management of memory containing sensitive data (such as keys or plaintext)
I would not expect a reviewer to consider the cryptographic strength of the algorithm. And if I were requesting such a review, I don't think I'd want to set great store in assurances given by strangers from the Internet...