About the role
What you will do:
- Build and run Kubernetes clusters on EKS, GKE and AKS with Terraform and Helm, including networking, IAM, secrets and storage.
- Package Matters for customer environments: managed Kubernetes, on premises RKE2 clusters, single VM installs and air gapped networks, through scripted, prompt driven installers.
- Own the release path: build and publish container images and Helm charts through GitHub Actions, gate releases on qualification tests, and run in place upgrades at customers without downtime.
- Operate the stateful services the product depends on: PostgreSQL and Aurora, Redis, RabbitMQ, object storage (S3, MinIO, Dell ECS), Temporal, and EFS or Filestore volumes.
- Set up and tune observability: metrics, logs, traces and alerting in CubeAPM, CloudWatch and Grafana, with alert rules that reach the right people before a customer notices.
- Harden environments: private endpoints, zero trust access through Twingate, least privilege IAM, secret rotation, OIDC SSO and network policies.
- Drive cloud cost work: rightsizing, off hours schedules, reserved capacity and per client cost attribution.
- Act as the technical contact for customer DevOps teams during installs and upgrades, including sizing checks before an install starts. Write the runbooks, client notes and upgrade guides that let the next engineer, or the customer, repeat what you did.
What we are looking for:
- 4 to 7 years in DevOps, SRE or infrastructure engineering, with at least 2 years running production Kubernetes.
- Strong on one major cloud (AWS preferred) and comfortable picking up a second. We run on AWS, GCP and Azure.
- Terraform in daily use: modules, remote state, plan review discipline and drift handling.
- Helm and Kubernetes internals: writing charts, and debugging scheduling, networking, storage, RBAC and autoscaling (HPA, KEDA).
- Linux depth on Ubuntu and RHEL family systems: systemd, networking, disks, package management, and shell scripts you would be happy to hand to a customer.
- CI/CD with GitHub Actions or similar, including container builds, registries and release gating.
- Hands on with Docker, containerd and at least one on premises Kubernetes distribution (RKE2, k3s or kubeadm).
- Databases, queues and object storage in production: PostgreSQL, Redis, RabbitMQ or Kafka, and S3 compatible stores.
- Observability as a practice: you have built alerting that people trust and have run incident response.
- Clear written English. Much of this job is runbooks, pull requests and messages to customer engineers you will never meet in person.
- Careful by habit. You back up before you change, you test on a staging VM before you share a command, and you label every step with where it runs.
Nice to have:
- Air gapped or offline delivery: image bundling, private registries such as Harbor, Zarf or Flux.
- AKS or GKE in production, beyond AWS.
- Running GPU or ML workloads on Kubernetes, including queue depth based autoscaling.
- Temporal or another workflow engine in production.
- Security tooling and audits: Cloud Armor or WAF, CIS benchmarks, SOC 2 or ISO 27001 evidence collection.
- Python or Go for tooling beyond shell.
- FinOps: AWS Cost Explorer, Savings Plans, tagging strategies and chargeback reports.
- Prior work at a startup shipping software into enterprise customers' own infrastructure.