Systems Engineer, Kernel (Automation/Validation)

CoreWeave is The Essential Cloud for AI™. Built for pioneers by pioneers, CoreWeave delivers a platform of technology, tools, and teams that enables innovators to build and scale AI with confidence. Trusted by leading AI labs, startups, and global enterprises, CoreWeave combines superior infrastructure performance with deep technical expertise to accelerate breakthroughs and turn compute into capability. Founded in 2017, CoreWeave became a publicly traded company (Nasdaq: CRWV) in March 2025. Learn more at www.coreweave.com . Senior Systems Engineer, Kernel Automation Validation CoreWeave is seeking a specialized Kernel Automation Engineer to join our HAVOCK Team. In this role, you will be the architect of our validation infrastructure. You won't just run tests; you will design the framework that ensures stability across the entire vertical stack—from Firmware to Kernel to NCCL to Slurm. We need an engineer who understands that a kernel is only as good as the workloads it supports. You will build the CI/CD pipelines and hardware-in-the-loop test harnesses that validate our custom Linux builds against next-generation GPUs, high-speed interconnects, and AI schedulers. H ardware - A cceleration - V irtualization - O perating Systems - C ontainerization - K ubelet Our Team’s Stack: Languages: Python (expert level), Go, Bash, C CI/CD Automation: GitHub Actions, Jenkins, Ansible, Terraform, ArgoCD Validation Tools: LAVA (Linaro), LabGrid, kselftest, Pytest, Syzkaller The Stack: BIOS/BMC Firmware, Linux Kernel, OFED/RDMA drivers, NCCL, Slurm Hardware: Bare metal provisioning (PXE, IPMI), NVIDIA GPU’s, ConnectX Focus Areas: End-to-End Test Framework: building a unified test harness that triggers validation from low-level firmware up to high-level Slurm job scheduling. Hardware-in-the-Loop: Managing and automating a fleet of bare-metal DUTs (Device Under Test) for kernel regression testing. Performance Validation: Automating benchmarks for NCCL (GPU-to-GPU communication) and RDMA to catch latency regressions before production. Build Pipeline Evolution: Refactoring the "Havock" build system to allow any engineer to spin up a custom kernel and test it on real hardware via CLI. Firmware/Driver Interop: Automating matrix testing between different Kernel versions, NVIDIA Drivers, and NIC Firmware. Responsibilities: Design and implement a "full-stack" validation pipeline that ensures a kernel change doesn't break NCCL rings or Slurm scheduling logic. Develop a CLI tool (Go/Python) that allows kernel engineers to launch ad-hoc test runs on specific hardware SKUs. Automate the provisioning of test nodes, including BIOS/BMC configuration and firmware flashing. Integrate standard kernel tests (LTP, kselftest, blktests) into a modern CI/CD workflow. Write "Glue" code to connect low-level kernel outputs (dmesg, serial logs) to high-level reporting dashboards. Requirements: 5+ years of experience in Test Automation, Release Engineering, or Systems Development. Strong Python/Go skills: Ability to write complex test frameworks and CLI tools (beyond simple scripting). The "Vertical" Perspective: Experience debugging or testing interactions between Firmware, Kernel, and Userspace. CI/CD Expertise: Experience building pipelines that manage hardware resources (not just software builds). HPC Knowledge: Familiarity with Slurm (job scheduling), MPI, or NCCL (collective communications). Experience with bare-metal provisioning (PXE, iPXE, Kickstart). Nice-to-haves: Experience testing RDMA/InfiniBand networks. Familiarity with NVIDIA GPU architecture and the CUDA ecosystem. Experience writing Reproducers for kernel bugs in C. Knowledge of PyTorch distributed training flows. Our compensation reflects the cost of labor across several US geographic markets. The base pay for this position ranges from $153,000/year in our lowest geographic market up to $242,000/year in our highest geographic market. Pay is based o ... (truncated, view full listing at source)