Track Lead (Support & Operations) (IN)
HCLTech
HCLTech
Join our team as a Track Lead focused on Support & Operations, contributing to cutting-edge GPU cluster networking.
This role demands deep expertise in data center fabric design and deployment, particularly with BGP and VXLAN-EVPN technologies. You will be instrumental in ensuring the seamless operation of high-performance computing environments.
Familiarity with InfiniBand, RDMA/RoCEv2, and lossless Ethernet tuning is essential. You will also gain exposure to NVIDIA's advanced networking platforms, including Spectrum-X Ethernet and BlueField DPUs.
Your understanding of Linux, including Cumulus Linux, and GPU cluster network design based on DGX architectures will be key. You will also be involved in fabric monitoring and troubleshooting, optimizing performance for 400G/800G optics and cabling.
Architect, deploy, and troubleshoot large-scale production clusters with a focus on NVIDIA DGX and GPU administration. You will leverage your understanding of AI training and inference to optimize network performance.
Demonstrate deep expertise across multiple network operating systems like Cumulus Linux, SONiC, IOS, Junos OS, and EOS. You will be responsible for advanced network concepts including congestion control, traffic engineering, telemetry, and performance optimization.
Drive automated provisioning using scripting languages like Python, and tools such as Ansible and Salt. Proficiency in CI/CD pipelines and API integration is expected.
Develop High-Level Design (HLD) and Low-Level Design (LLD) documents, validating designs before production rollout. A strong grasp of the interplay between networking, GPU computing, distributed storage, and HPC workloads is crucial.
A strong foundation in BGP (eBGP underlay, BGP unnumbered, ECMP) for leaf-spine/Clos data center fabrics is required. Experience with VXLAN-EVPN overlays, IRB, and multi-tenancy using VRFs is also essential.
Conceptual knowledge of InfiniBand, RDMA/RoCEv2, and lossless Ethernet tuning (PFC, ECN, DCQCN) is necessary. You should be comfortable picking up new Linux distributions like Cumulus Linux quickly.
Proven experience with Linux and Kubernetes is vital. A good working understanding of High-Performance Computing (HPC) environments is also a must.
Proficiency in fabric monitoring and troubleshooting using streaming telemetry and Linux networking tools, extending to high-speed optics and cabling, is expected.
HCLTech
IT Consulting