It is hard, but definitely possible with the right motivation. The Gaussian correlation conjecture was a major open problem in Probability and convexity, until it was solved by https://en.wikipedia.org/wiki/Thomas_Royen, a retired statistician.
And a very different example: If you search for the publications of James Simons at Mathscinet (which are arranged in reverse chronological order), you will find the following two adjacent publications, with only one 1992 interview noted in the intervening 23 years:
[1] Simons, James; Sullivan, Dennis Axiomatic characterization of ordinary differential cohomology. J. Topol. 1 (2008), no. 1, 45–56.
[2] Cheeger, Jeff; Simons, James Differential characters and geometric invariants. Geometry and topology (College Park, Md., 1983/84), 50–80, Lecture Notes in Math., 1167, Springer, Berlin, 1985. (Reviewer: P. Molino)
The 2008 paper was followed by a number of other significant joint papers by Simons and Sullivan.