Runtime Engineer

What you'll do

• Develop device memory, DMA, streams, events, synchronization, executable formats, and custom-kernel ABI and marshaling.
• Build Python bindings and LLM inference serving and cluster orchestration, including scheduling, token streaming, and recovery.
• Design profiler and debugger interfaces and meet runtime-overhead and serving-throughput targets.

What they ask for

• Strong systems-language experience (Rust, C, C++, or Go), including memory management, allocator design, and FFI/ABI work.
• Production Python interop experience (PyO3, ctypes, pybind11, or equivalent C-ABI bridging).
• Experience designing and maintaining versioned API or ABI contracts between teams.
• Hands-on experience with an accelerator programming model, plus familiarity with ML systems, training and inference, collectives, and tensor layouts.

Nice to have

• LLM inference internals such as vLLM, TensorRT-LLM, or SGLang.
• Rust at depth, custom allocator or low-level memory work, ML framework integration, profiling/tracing, driver-adjacent work, or new-silicon bring-up.

Pay

$160,000 - $250,000 USD base salary; $175,000 - $362,500 USD base salary; $250,000 - $475,000 USD base salary

Tools

Rust, C, C++, Go, PyO3, DLPack, CUDA, ROCm

AI focus

Build runtime and LLM inference serving software for custom silicon used for large-language-model inference and training.