Job Description & Details
This is a heavy-duty systems engineering role focused squarely on squeezing every last drop of performance out of ARM bare-metal compute hosts through hardware offloading and interconnect optimization. If you enjoy diving deep into low-level kernel code, profiling PCIe bottlenecks, and working with cutting-edge acceleration tech, this 24-month contract gives you a massive canvas to make a real impact.
What You'll Actually Be Doing
You will spend your days implementing and optimizing hardware-based network offloading like SR-IOV and NFV, alongside integrating and benchmarking NVLink support for CPU-to-CPU interconnect performance. Expect to spend significant time profiling hardware bus bottlenecks, analyzing PCIe throughput, and examining latency characteristics on ARM superchip platforms. You'll be writing both kernel and user-space code enhancements to push platform throughput to its absolute limits.
The Core Tech Stack
You need absolute mastery of C and C++ for bare-metal performance engineering, coupled with deep, production-level background in offload technologies like SR-IOV, NFV, and PCIe device pass-through. Familiarity with kernel bypass frameworks like DPDK, eBPF, or direct memory access (DMA) drivers is non-negotiable because the team expects you to jump straight into the deep end of high-speed interconnects and ARM superchip optimizations.
Interview Expectations
You will likely be asked to explain how you would debug a mysterious latency spike in a PCIe device pass-through setup under heavy load, so be ready to walk through your exact profiling methodology using standard Linux performance tools. The hiring manager is secretly looking for someone who doesn't just know the theory of kernel bypass, but has actually wrestled with DPDK or eBPF in production and understands the memory barriers and cache coherency issues involved.
Application Advice
Make sure your resume screams 'low-level systems and performance tuning' rather than general application development. Put C, C++, SR-IOV, NFV, PCIe, DPDK, and NVLink right at the top, and explicitly detail a past project where you optimized hardware throughput or resolved a nasty kernel-space bottleneck to easily clear the ATS filters.