1 match
Write CUDA and low-level kernels and optimizations for vLLM on NVIDIA GPUs and cross-platform accelerators, working with hardware vendors on new chips. Requires CUDA or equivalent (CuTeDSL, Triton, TileLang), C++ and Python, and profiling with Nsight or rocprof. Preferred experience includes FlashAttention-style fused kernels and quantization (INT8, FP8).
Inferact's mission is to grow vLLM as the world's AI inference engine and accelerate AI progress by making inference cheaper and faster. Founded by the creators and core maintainers of vLLM, we sit at the intersection of models and hardware—a position that took years to build.
We're looking for a kernel engineer to squeeze every FLOP out of modern accelerators. You'll write the kernels and low-level optimizations that make vLLM the fastest inference engine in the world. Your code will run on hundreds of accelerator types, from NVIDIA GPUs to emerging silicon. When hardware vendors develop new chips, they integrate with vLLM. You'll work directly with these teams to ensure we're extracting maximum performance from every generation of hardware.
Inferact has three kernel roles. If your kernel work is mostly on Google TPUs (Pallas, JAX), apply to TPU Kernel Engineering. If it is mostly on AMD GPUs (HIP, CK, AITER), apply to AMD GPU Kernel Engineering. For NVIDIA GPUs and cross-platform kernel work, apply to Kernel Engineering. Apply to one; we route you to the best fit.
Minimum qualifications:
Preferred qualifications:
Bonus points if you have:
Inferact, Inc. is an E-Verify company, and you may review the Notice of E-Verify Participation and the Right to Work posters in English and Spanish.
Compensation: $200K - $400K
