When data quality is fine, there are two usual suspects: data drift or concept drift. Data Drift -- The input data has changed. The distribution of the variables is meaningfully different. As a result, the trained model is not relevant for this new data. Concept Drift -- In contrast to the data drift, the distributions might even remain the same.

Actually, whatever your job is, Laurent Alexandre once said "If you AI AI, you will soon be unemployed". So if you are NOT complementary to an AI ... So far, these are the best public datasets to feed your ML, DL, AI models. Google and NASA's Quantum Artificial Intelligence Lab appears to be Google and NASA's answer to your question. Software engineer by day and a geek by night, I've made a not-few projects using various programming languages mainly Java & JS). Just start and learn things as long as you build your apps.