Time varying (dynamic) data sets for practice

I just finished some courses on machine learning in production that teaches the concepts and techniques used in training, serving and maintaining ml models. Now I want to do my own project the apply the things I learn. I need a dynamic structured dataset that changes over time to make the need for large scale deployment and induce model decay. I am also thinking about simulating data and concept drift with some noise so if anyone has something that helps with such problem please help.