Description
Meta is hiring a senior kernel and performance engineer for its MTIA Software team to design, implement, optimize, and deploy high-performance compute and communication kernels for AI accelerator ASICs. The role covers GEMM, attention, normalization, collectives, quantization, numerics, kernel authoring frameworks, pre-silicon validation, hardware/software co-design, and production deployment across PyTorch workloads. It requires a bachelor's degree, at least six years of relevant experience, strong C++ and Python skills, parallel-architecture kernel experience, and a measurement-driven performance approach.
