Jobs

Inferact is seeking a TPU Performance Engineer to join their team in Singapore, focusing on optimizing vLLM for Google TPUs. This role involves building and optimizing TPU backends, compiler integrations, and benchmarking infrastructure using technologies like JAX, XLA, and Pallas. The ideal candidate will have a deep understanding of TPU execution and experience optimizing ML kernels and inference paths. You will work at the intersection of inference systems, compilers, and hardware architecture to deliver frontier inference performance and ensure TPU support in vLLM is fast and maintainable.