Accelerates ML model training and inference across CPU, GPU, NPU hardware. Cross-platform engine supporting 6+ languages.
