System Engineer (Token Factory)

About Nebius:

Nebius is leading a new era in cloud infrastructure for the global AI economy. We are building a full-stack AI cloud platform that supports developers and enterprises from data and model training through to production deployment, without the cost and complexity of building large in-house AI/ML infrastructure.

Built by engineers, for engineers. From large-scale GPU orchestration to inference optimization, we own the hard problems across compute, storage, networking and applied AI.

Listed on Nasdaq (NBIS) and headquartered in Amsterdam, we have a global footprint with R&D hubs across Europe, the UK, North America and Israel. Our team of 1,500+ includes hundreds of engineers with deep expertise across hardware, software and AI R&D.

About the role:

Token Factory is a part of Nebius Cloud, one of the world’s largest GPU clouds, running tens of thousands of GPUs. We are building an inference platform that makes every kind of foundation model — text, vision, audio, and emerging multimodal architectures — fast, reliable, and effortless to deploy at massive scale.

Responsibilities:

Develop and optimize low-level kernels and runtime components for AI inference

Improve performance of inference engines GPU platforms

Profile and debug system-level and hardware-level performance issues

Integrate support for new hardware architectures (Hopper, Blackwell , Rubin )

Collaborate with ML and backend teams to optimize end-to-end execution

Required Qualifications:

Strong proficiency in C++ , OR expertise in GPU programming with a focus on low-level high-performance coding and memory management

Experience in GPU programming or systems-level software development , e.g. operating system internals, kernel modules, or device drivers

Hands-on experience with profiling and debugging tools to identify performance issues on both CPUs and GPUs, and the ability to optimize code based on those findings.

Solid understanding of CPU/GPU architecture and memory hierarchy

Preferred Qualifications:

Experience with GPU computing programming : CUDA, ROCm , CUTLASS, Cute, ThunderKittens , Triton, Pallas, Mosaic GPU

Familiarity with ML inference runtimes (e.g. TensorRT , TVM)

Knowledge of Linux internals, drivers, or compiler toolchains

Experience with tools like perf, VTune , Nsight, or ROCm profiler

Familiarity with popular inference engines (e.g. such as vLLM , sglang , TGI)

We conduct coding interviews as part of the process.