During the SciPy 2020 conference (held virtually that year), I was digging through the project improvement proposals and found a suggestion for a faster convex hull algorithm. A bit more digging uncovered a submission to fulfil this proposal, but it had not been accepted due to a bug. During the rest of the sprint I debugged how the error in question had arisen, and proposed a fix in a new pull request.
WIP/review underway