Staff GPU Software Engineer
LOCATION: REMOTE IN THE US
Who We Are:
GigaIO has invented the first truly composable universal dynamic infrastructure fabric, empowering users to accelerate workloads on-demand, using industry-standard PCI Express/soon CXL technology. As a global leader in Composable Infrastructure (CI), our open platform allows users to effortlessly manage and configure “physical” bare metal server systems in seconds. GigaIO’s FabreX™ open architecture allows users to configure their physical server infrastructure on-the-fly while enabling true datacenter disaggregation.
Does getting in on the ground floor of a data center technology that is disrupting the computing, HPC, and AI industry make your heartbeat a little faster? Does the excitement of joining a team of exceptionally talented and motivated technologists at a well-funded, startup sound attractive?
What You Will Do:
We are looking for a GPU Software Engineer to help build our software platform and infrastructure to leverage our FabreX™ interconnect for best-in-class performance for GPU accelerator utilization in HPC applications, including GPU-to-GPU and GPU-to-Storage communication. The successful candidate will be accustomed to developing High-Performance Linux Drivers and Libraries, and take initiative and ownership for extracting the highest performance possible from the unique capabilities available in FabreX™. The candidate will work with engineering leadership to establish objectives, and execute strategic, cross-functional initiatives to solidify and evolve the operating system driver foundation for NVIDIA, Intel and AMD GPU computing hardware and software products.
- 5+ years of development experience in systems software
- Good understanding of CUDA, OpenACC, OpenCL, GPU Direct and GPU Direct RDMA, NCCL/RCCL.
- Knowledge of Linux drivers and kernel space programming
- Strong Linux architecture knowledge
- Proficiency in C
Qualification and Experience:
- Experience writing Linux device drivers for high-performance I/O devices, such as NVMe, OFED (InfiniBand, RoCE, iWARP) or DPDK.
- Experience with PCI Express and single-root I/O virtualization (SR-IOV) a plus
- Experience with open source community initiatives a plus
A clear passion for learning emerging technology independently and BS degree (MS or PhD desired) in Computer Engineering, Computer Science or similar a plus