Location: Indore
Employment Type: Full time
About RackBank Datacenters:
RackBank Datacenters is a fast-growing and trusted data center company committed to powering digital transformation with secure, scalable, and efficient IT infrastructure solutions. With state-of-the-art data centers in India and a strong client base across multiple industries, RackBank is focused on delivering next-gen cloud and colocation services.
Our Vision:
To build India’s most reliable and trusted data center and digital infrastructure ecosystem, empowering enterprises and governments with secure, scalable, and future-ready infrastructure.
About the role:
NeevCloud runs production Kubernetes clusters for its GPU cloud platform, with Ceph as the storage backend and a mixed fleet of NVIDIA datacenter and workstation - class GPUs. You will own the day - to - day health of these clusters and their nodes: Keeping them stable, patched and monitored, and fixing issues before customers notice them. As the platform grows, you will also deploy Kubernetes on new GPU Clusters.
What you will do:
Cluster Operations:
- Run daily health checks on the production clusters: nodes, control plane, etcd, certificates and capacity.
- Plan and carry out Kubernetes version upgrades with a rollback plan and minimal downtime.
- Troubleshoot pod failures, evictions (disk and memory pressure), scheduling problems, and networking and DNS issues, including customer pods that carry a direct public IP.
- Support VM workloads running on KubeVirt alongside containers.
- Manage persistent storage through Ceph CSI (RBD and CephFS): PVC issues, capacity and performance.
- Watch Ceph cluster health with the storage team, and act on warnings that affect Kubernetes volumes (slow requests, nearly full OSDs, degraded placement groups).
- Maintain monitoring and alerts in SigNoz and Zabbix, and respond to alerts.
- Take etcd backups, test restores, and keep disaster recovery steps documented.
- Manage RBAC, namespaces, resource quotas and limits, and work with the security team on vulnerability fixes.
Node Maintenance:
- Handle the node lifecycle: add and remove worker nodes, and cordon, drain and return nodes after maintenance.
- Patch the OS and kernel, and upgrade GPU drivers and firmware in a planned window, with the data centre team where needed.
- Use out-of-band access (BMC, IPMI or Redfish) for remote console, power actions and hardware logs.
- Find hardware faults (disk, memory, NIC, GPU), raise vendor tickets, and work with the data centre team on replacements.
- Find GPU faults using nvidia-smi, DCGM metrics and Xid errors, and decide whether a node needs draining, a driver fix or a hardware ticket.
- Check every node before it returns to production: GPU health, driver and firmware versions, network and storage access.
New GPU Cluster builds:
- Deploy Kubernetes on new GPU node clusters, from bootstrap to production handover.
- Set up the NVIDIA GPU Operator and Network Operator, including RDMA networking (RoCE or InfiniBand) for multi-node workloads.
- Validate new clusters with GPU burn-in and NCCL tests before handover.
- Automate node and cluster setup (Ansible or similar) so every build is repeatable and documented.
Support and on - call:
Take escalations from the L1 support team, write the root cause analysis for incidents, and keep runbooks up to date.
Join the on-call rotation for production incidents.
What we need:
- 4+ years of Linux administration (Ubuntu primary; RHEL a plus): systemd, kernel, networking, storage and logs.
- 2+ years of hands-on work running Kubernetes in production, not only lab or course work.
- Has built at least one Kubernetes cluster from scratch (kubeadm, Kubespray or similar).
- Confident troubleshooting with kubectl: logs, events and describe output.
- Working knowledge of Helm, at least one CNI (Cilium, Kube-OVN, Calico or similar), and ingress.
- Ceph: understanding of the architecture (MON, OSD, MGR, pools, placement groups, replication and erasure coding), reading ceph -s and health warnings, and using RBD and CephFS through the Ceph CSI driver.
- GPU: NVIDIA GPU Operator, driver and CUDA version compatibility, GPU scheduling in Kubernetes, and GPU monitoring with DCGM.
- Server hardware basics and out-of-band management (BMC or IPMI).
- Ansible or a similar automation tool, plus scripting in Bash or Python.
- Experience with monitoring and alerting tools (SigNoz, Zabbix, Prometheus or similar).
Good to have:
- CKA or CKS certification.
- KubeVirt administration (live migration, PCI passthrough).
- NVIDIA Network Operator, RDMA (RoCE or InfiniBand) and NCCL testing.
- GPU time-slicing or MIG, and awareness of NVLink and PCIe topology.
- Rook, Ceph upgrades or OSD replacement.
- GitOps tools (Argo CD or Flux), or policy tools such as Kyverno.
- Experience in a data centre or GPU cloud environment.
- We operate with:
- Freedom and responsibility
- High performance standards
- Honest, direct communication
- Respect for context over process
Interested?
If ownership, impact, and building India’s digital infrastructure excite you, apply now.