events

Kernel Engineering: From CUDA Foundations to Frontier AI Kernels (AIPERF15 15%)

Vizuara runs a live online Kernel Engineering cohort in partnership with Crusoe. Participants learn how GPU kernels work from first principles, then write, profile, and optimize them in practical coding sessions. The curriculum spans parall

In inglese

Starts

2026-10-13T01:30:00+00:00

Organizer

AI Performance Engineering Meetup (San Francisco, Global)

Location

Online event

Vizuara runs a live online Kernel Engineering cohort in partnership with Crusoe. Participants learn how GPU kernels work from first principles, then write, profile, and optimize them in practical coding sessions. The curriculum spans parallelism, the roofline model, CUDA, GPU memory, matrix multiplication, tensor cores, and attention kernels, ending with modern tools and an engineering capstone. Recordings and code are included. It is designed for engineers comfortable with Python and basic PyTorch; no prior CUDA or C++ experience is required.