Orcrist Technologies - Infrastructure/Systems Engineer

• Design, size, provision, and operate bare-metal server fleets across on-prem and air-gapped environments (firmware/BIOS/UEFI, BMC via Redfish/IPMI, OS, RAID, kernel and storage tuning) using zero-touch provisioning (PXE/iPXE, MAAS/Metal3/Tinkerbell/Ironic) and automation (Ansible, Terraform, or equivalent — the tooling matters less than the automation mindset). • Build and run modern data-center networking: L2/L3 design, IP Fabric, BGP and switching, and RDMA fabrics (RoCE/InfiniBand) sized to scale without ripping out the core. • Engineer resilient, highly available storage (Ceph/Rook, NVMe) with capacity planning and encryption at rest. • Operate confidently in air-gapped and on-prem environments: offline mirrors and registries, signed artifacts, firmware/driver lifecycle without internet access, and system hardening. • Support MLOps and inference-serving fundamentals — GPU model serving (Triton/KServe/vLLM), GPU scheduling and sharing, and throughput/latency optimization — in partnership with our SRE and ML teams. • Plan and run on-site build-outs: rack integration, power budgets, thermal/cooling and UPS sizing, commissioning, capacity planning, runbooks, and operator handover, with SWaP awareness for field sites. • 5+ years in bare-metal, data-center, or systems infrastructure engineering, with hands-on ownership of physical and compute infrastructure at scale. • Strong bare-metal Linux (Ubuntu, RKE2, Talos, or similar): provisioning, firmware/BMC management, PXE/iPXE, kernel and storage tuning, systemd, RAID. • Real experience with infrastructure automation (Ansible, Terraform, or equivalent), Git and CI/CD, and scripting in Python, Bash, or Go. • Solid, current data-center networking fundamentals: L2/L3, IP Fabric, BGP, and switching — this is a hard requirement, not a nice-to-have. RDMA (RoCE/InfiniBand) experience is a strong plus. • Comfortable operating in air-gapped or on-prem environments and traveling to customer sites for builds and deployments. • Practical hardware sizing literacy: power budgets, thermal/cooling, UPS sizing, and rack integration. • Documentation-focused, methodical, and calm during hardware incidents. Eligible to work in Germany. • Nice‑to‑haves • German language (B1+); exposure to regulated or security-critical environments (e.g. BSI C5, ISO 27001, or defense-sector delivery) is a plus. • NVIDIA GPU stack knowledge (drivers, CUDA, GPU Operator, MIG, DCGM) and cross-node GPU interconnect experience (NVLink, InfiniBand, NCCL). • Kubernetes bare-metal fundamentals — how cluster bring-up and GPU device plugins interact with the underlying hardware, not day-to-day cluster operation. • Inference optimization (vLLM, TensorRT-LLM, quantization) and familiarity with switch NOS (SONiC/Cumulus). • Relevant certifications (NVIDIA, Red Hat, CKA/CKS) or field/forward-deployed engineering experience.