Referencing such a document is a well-know proof technique, often stated as follows (cf. proof techniques):
Proof by reference to inaccessible literature:
The author cites a simple corollary of a theorem to be found in
a privately circulated memoir of the Slovenian Philological Society, 1883.
In your case, the quality of the work seems to be a non-issue due to the reputation of the mathematician, but what is an issue is the long-term availability of the document.
I ran into references to unpublished works a few times. For a particular example in my mind, everyone could find it, apparently, several decades ago, but now the document is gone. I must say I really hate it.
So, provide the reader with all the information you can find that would help them to get the manuscript. E.g.:
Abra K. Dabra, On the importance of being published, 1883, Slovenian Philological Society, unpublished technical report, retrieved from the author's Web page http://www.example.com on 2018-02-30.
or
Abra K. Dabra, On the importance of being published, 1883, Slovenian Philological Society, unpublished technical report, available from the library of the Slovenian Philological Society, street house-no, 1000 Ljubljana, Slovenia.
or
Abra K. Dabra, On the importance of being published, 1883, Slovenian Philological Society, unpublished, retrieved from jstor.org/... on 2018-02-30.
etc.
In addition, as @darij-grinberg says, you may ask the author to store his/her paper with an independent service for long-term archival purposes.
If you reference hard claims with their proofs, you might wish to copy the material to the appendix of your document to make your document self-contained. Of course, get the permission from the original author first, acknowledge him or her properly, and mark the corresponding part as a citation.