You can pick any challenge around time-series, bring the data to TimeNet, and then train or fine-tune an ML model on it.
We're providing compute credits and prizes (AWS, GCP & Nebius).
First deadline: November 22