Multiple ones.

• Wave Function Collapse for an existing tileset (Warcraft 2) - it's really hard to figure out how to build the data on whether a tile fits on a certain side of another tile. From the top of my head: Some 180 tiles total * 4 sides to compare * 180 options per side = 129_600 total checks I would need to do, which is just wayyy too much. Need to figure out how to reduce these checks (or automate them using OpenCV or something - not sure yet)
• Comparing dual datasets of people who worked on certain games (the games being all Warcraft games), so I can see who worked on Game A, who on Game B, and who on both. Am currently (manually) cleaning up the data, because Blizzard has been very inconsistent in how they structure their credits. At least WoW had everything available in HTML - having to manually copy over 600+ people would be no fun (per expansion).
• A cryptocoin predictor for my friends and I. One part ingestion from an API into a Postgres DB, one part Streamlit + some stats / data science stuff that I'm very much a beginner in. The platform we're on doesn't provide much predictive power. Using the typical Simple Moving Averages and the like simply doesn't predict clearly enough, IMO.

Oh, and I'm also setting up a private ProxMox server, over at a friends house, and need to connect that to the NAS over at my house, so we can copy over our data for redundancy/backup purposes.

I'm bouncing between these projects.