Staff Engineer, Bare Metal Kubernetes Platform

Equinix is the world’s digital infrastructure company®, shortening the path to connectivity to enable the innovations that enrich our work, life and planet. 
 

A place where bold ideas are welcomed, human connection is valued, and everyone has the opportunity to shape their future.

Job Summary

We are looking for a Staff Engineer to join the Reliability Engineering team that operates a bare-metal Kubernetes platform across multiple metro locations. The platform's architecture is defined and the build is underway you will provision and operate it day to day: bringing up nodes and clusters, running upgrades, keeping the fleet consistent through GitOps, and owning your shift of the on-call rotation.

This is a deliberately broad role. We are not looking for someone who works one layer and hands everything below it across a boundary. You will work alongside the teams that own the physical hardware, the underlay network, and the data plane software, and you need working knowledge of all three enough to isolate where a problem lives before escalating it.

You will join the Digital Interconnection Engineering organization. The team owns the full application stack for interconnection and the Kubernetes platform for the software-defined data plane across multiple metros. We operate with an automation-first, GitOps-driven culture and a strong commitment to operational rigor, documented runbooks, and blameless postmortems.

Responsibilities

OS Provisioning & Bare-Metal Infrastructure

• Provision bare-metal servers across multiple metros using the team's PXE/iPXE imaging and cloud-init pipeline, and harden the OS prior to cluster bootstrap

• Extend and maintain that provisioning automation as new hardware, metros and OS versions are introduced, keeping node bring-up free of manual per-node steps

• Work with the teams that own the hardware on BIOS/firmware baselines, NIC configuration and disk layout standards, and use BMC/IPMI and platform sensors to isolate hardware faults before handing them over

Kubernetes Cluster Operations

• Bring up and operate Kubernetes clusters following the team's established control-plane topology, including etcd members, kube-apiserver HA behind the L4 load balancer or VIP, and defined failure-domain placement

• Provision worker nodes, configure CNI, and run end-to-end cluster validation before workloads are onboarded

• Perform routine cluster operations: node drains and replacements, certificate rotation, capacity checks, and remediation of cluster-level alerts

GitOps & Fleet Configuration

• Make cluster and fleet configuration changes through GitOps, keeping clusters consistent, auditable and drift-free

• Perform cluster registration, RBAC and policy changes in the management plane, respecting the isolation between production and nonproduction planes

• Extend the deployment pipeline for the software-defined data plane adding validation coverage and test cases working from the pipeline design and gates owned by the senior engineers on the team

Upgrades & Patching

• Execute rolling OS and Kubernetes upgrades across the fleet from tested runbooks, with no unplanned downtime, and improve those runbooks as you find gaps

• Take etcd snapshots ahead of significant changes, and run scheduled restore drills to prove the recovery path works

• Apply the platform's patching cadence across the fleet, tracking CVEs affecting the OS, Kubernetes and cluster components

Observability & On-Call

• Own your shift of a follow-the-sun on-call rotation, and act as first responder for control-plane, node, and cluster-level incidents

• Build and tune dashboards and alerts for the systems you operate, and reduce alert noise where it is costing the rotation time

• Write up incidents you respond to, contribute to blameless postmortems, and turn recurring toil into automation

• Author and maintain runbooks for the components you operate

What Success Looks Like First 9-12 Months

• Provisioning and bringing up nodes and clusters independently, without per-node manual steps

• Running OS and Kubernetes upgrades on cadence from runbooks, and improving those runbooks where they fall short

• Owning your on-call shift unassisted, as first responder for cluster and node-level incidents

• Restore drills run on schedule, with the recovery path demonstrated rather than assumed

• Measurable reduction in toil or alert noise in the areas you operate

• Runbooks and dashboards in place for the components you own

Qualifications

Required

• Linux on bare metal: Solid hands-on Linux/Ubuntu administration and provisioning; familiarity with PXE/iPXE imaging or equivalent, cloudinit, and OS hardening

• Hardware working knowledge: Comfortable with BMC/IPMI consoles, SMART data and platform sensors; able to identify a failing disk, NIC, DIMM or power supply and follow a vendor replacement process

• Kubernetes: Production experience operating clusters ideally RKE2 or kubeadm on bare metal including etcd basics, node lifecycle, CNI configuration and cluster-level troubleshooting

• Infrastructure as code: Day-to-day use of Ansible, Terraform, ArgoCD, Fleet or similar; comfortable making infrastructure changes through a pipeline rather than by hand

• Networking: Working knowledge of VLANs, BGP, bonded NICs and SR-IOV enough to read a configuration, interpret what you see, and isolate whether a problem is above or below the network layer

• Upgrades: Experience executing OS and Kubernetes upgrades in production against an established runbook

• On-call: Experience in an on-call operations role, responding to production incidents and writing them up afterward

• Automation mindset: A reflex to automate repeated manual work, and to leave the runbook better than you found it

Preferred

• Experience with Rancher, Rancher Prime or a comparable multi-cluster management plane

• Familiarity with DPDK-based data planes such as 6WIND VSR, or with SR-IOV NIC tuning

• Exposure to SDN or private-connectivity platforms (Equinix Fabric, VyOS, or equivalent)

• Familiarity with LLM provider APIs and AI/agent gateway proxying patterns, since inference traffic from multiple providers transits this platform

• Experience operating infrastructure across geographically distributed sites