Automating a high-touch financial process in Python at Angel Studios
I had the pleasure of working on a really fun project at Angel Studios called Disbursements. It cut a month-long quarterly process into something that can be run in a few hours, and saved about $100,000 annually in wages spent on running the process. This was a dreaded financial process in which we pay creators on a quarterly cadence to meet our contractual obligations and maintain a relationship with them. We get revenue from different platforms, we split that revenue up by creator, and then we pay the creators for the revenue they earned. Sounds simple enough. However the reality was that the process has been plagued by problems for years. It took a data scientist a full month every quarter to process the data and generate reports. Reports were often wrong or incomplete - creators would often complain that their historical earnings went down, for example. Reports were also not transparent, so the creators would have no way to know if calculations were done accurately or not. Combinin...