Arcane ML

Production-ready distributed ML training framework across SSH clusters, Modal Cloud GPUs, and local multi-GPU setups.

Distributed training framework with PyTorch DDP, automatic gradient synchronization, and a unified CLI abstracting SSH clusters, Modal Cloud GPUs, and local multi-GPU workflows.