Another thing I would like to mention is that in the last 5 years in Meta, I always face the situation that as a manager/director, I cannot be too hands-on to take over my reports' scope.
I was warned in the performance review at some point. After the OpenGo project in which I wrote 90% of the code, I was told "researchers need to do research and engineers do engineering work".
Therefore, doing first author/solo work is one way to keep my sense of owning things end to end, and get to know what's really going on.
Now I am super happy that the constraint is gone forever, thanks for the AI coding tools.
Congratulations! I always admire senior authors that have time and energy to do solo work.
Would you mind sharing a bit on how you balance this type of work with joint work with others? What are the main challenges?