Learn about carbon-efficient resource scheduling for machine learning clusters in this 17-minute conference talk from NSDI '25. Explore the problem of scheduling ML jobs while reducing carbon emissions, moving beyond traditional cluster schedulers that focus solely on job completion time optimization. Discover GREEN, a novel ML cluster scheduler that balances both time-efficiency and carbon-efficiency through a unique carbon-aware scheduling algorithm. Understand how GREEN leverages the temporal flexibility of ML jobs to shift workloads to less carbon-intensive times while maintaining overall daily capacity. Examine experimental results using real ML job workloads that demonstrate up to 41.2% reduction in cluster-wide carbon footprint and 12% reduction in peak power consumption, with only 3.6%-5.9% time efficiency tradeoff compared to existing methods.