Normal Computing
Runtime Engineer
$200K–$300K
Mountain View +1 · On-site · Full-time · Posted Oct 20, 2026
- Location
- Mountain View
New York
About the job
Own the architecture and implementation of a runtime that loads, schedules, and executes compiled programs on Normal Computing’s accelerator hardware, initially using simulators and FPGA prototypes.
What you'll do
- Design runtime components for device management, memory, executable loading, and command submission, and provide a stable interface for compiler and serving layers.
- Abstract hardware differences across simulators, FPGA prototypes, and silicon; manage device-resident state and keep runtime overhead low for inference.
- Set technical direction and define interfaces with compiler, serving, driver, and firmware teams.
What they ask for
- Experience building runtimes, drivers, or hardware abstraction layers for accelerators or other non-standard hardware.
- Strong C++ or Rust systems programming skills, including memory management, concurrency, and low-level performance.
- Track record of owning a complex systems component end to end, from design through production, and driving technical decisions across teams.
- Understanding of LLM inference workloads and experience building software for evolving hardware, from simulation through bring-up.
Nice to have
- Experience with IREE's HAL, CUDA or ROCm runtime internals, in-memory compute, or other non-von-Neumann hardware.
- Experience integrating with an inference server such as vLLM, or contributing to open-source runtime or systems infrastructure.
Pay
$200K – $300K USD base salary
Tools
C++, Rust, IREE's HAL, CUDA, ROCm, vLLM
AI focus
Build runtime software for accelerator hardware that executes large-model inference, with performance measured by end-to-end latency and throughput.
Model-written summary · the original posting is authoritative