Discover how to scale NumPy operations across CPUs and GPUs with minimal code changes in this 25-minute PyCon US talk. Learn about cuPyNumeric, a drop-in replacement for NumPy that automatically parallelizes operations across distributed computing resources. Explore a real-world case study from SLAC National Accelerator Laboratory where scientists achieved 6x speedups for processing large experimental datasets during time-sensitive beam time operations. Understand how cuPyNumeric leverages Stanford University's task-based distributed runtime to handle data distribution, communication, and accelerated execution while maintaining compatibility with popular Python libraries like SciPy, matplotlib, and JAX. Gain insights into the implementation details and see demonstrations of both the productivity benefits and performance improvements this library offers for data and simulation scientists working with resource-intensive computations.