I presented PAL, which designs performance variability-aware schedulers for GPU-rich clusters, at SC’24 in Atlanta, GA.