About the company
This position is listed on behalf of a partner company, who manages all applications and next steps. Our partner is looking for a Staff Network Engineer (AI Fabric, Datacenter and Edge Networking) - Radian Arc based in Germany.
Responsibilities
- Design, deploy, and operate high-performance GPU networking fabrics for distributed AI workloads, including RoCE, RDMA, Spectrum-X, leaf-spine, fat-tree, rail, and multi-plane architectures.
- Analyze GPU communication patterns and optimize east-west traffic, congestion control, latency, throughput, and reliability for distributed training and inference workloads.
- Own Layer 2/3 architecture using BGP, ECMP, EVPN/VXLAN, VLAN/VRF, OVS/OVN, Linux networking, bridges, overlays, and scalable routing patterns.
- Design and operate north-south connectivity, WAF and application-layer protections, TLS termination, DDoS mitigation, and gateway infrastructure.
- Design private connectivity, dark-fiber and metro-fiber rings, high-capacity WAN links, DWDM transport, inter-site BGP routing, redundancy, and failure-domain isolation.
- Lead infrastructure initiatives from architecture and lab validation through BOM validation, datacenter layouts, deployment, acceptance, and production rollout.
- Establish automation for provisioning, configuration management, monitoring, lifecycle management, validation, and day-2 operations while improving reliability and recovery practices.
- Act as the senior escalation point for complex network incidents, leading investigation, root-cause analysis, remediation, and long-term architectural improvements.
- Define and monitor SLAs, SLOs, latency, reliability, capacity, and operational metrics, translating findings into concrete engineering improvements.
- Partner with infrastructure, platform, SRE, compute, storage, observability, and datacenter teams while setting networking standards and influencing long-term architecture.
- Serve as the primary networking design authority, mentor engineers in adjacent domains, and establish reusable patterns that can scale with the organization.
- Build Python and/or Bash tooling, validation frameworks, and operational systems that reduce manual work and increase consistency across deployments.
Requirements
- Strong hands-on experience designing and operating large-scale datacenter networks in production, with expert knowledge of BGP, OSPF, ECMP, and EVPN/VXLAN.
- Deep experience designing and operating high-performance GPU networking fabrics for distributed AI workloads, including practical RoCE/RDMA expertise.
- Strong understanding of NCCL communication patterns—including all-reduce, all-gather, broadcast, and reduce-scatter—and their implications for network topology and performance.
- Experience tuning large-scale RoCE fabrics, including PFC and ECN congestion management, and diagnosing NCCL stalls, RDMA congestion, fabric hotspots, and packet-loss issues.
- Proven experience with high-speed Ethernet and NVIDIA/Mellanox networking platforms, as well as high-performance interconnects, optics, and networking hardware.
- Ability to investigate complex cross-layer problems spanning hardware, firmware, Linux/kernel networking, and distributed application communication.
- Strong Python and/or Bash skills, with experience applying software engineering practices to infrastructure automation and building reusable operational tooling.
- Experience designing network observability and independently analyzing operational data, ideally using tools such as PostHog or equivalent systems.
- Demonstrated ability to own both long-term architecture and hands-on implementation in lean or rapidly scaling environments.
- Proven ability to lead complex initiatives, establish engineering standards, influence multiple teams without formal authority, and balance performance, reliability, scalability, operability, and cost.
- Strong stakeholder management and communication skills, with the ability to explain architectural decisions, trade-offs, and technical risks clearly.
- Ability to raise the technical bar of adjacent engineering teams and contribute to the development of a future networking organization.
- Professional working proficiency in English; the ability to collaborate effectively in a global team.