Why this role exists

Simplyblock builds Kubernetes-native, software-defined NVMe-over-Fabrics block storage. Our customers run it underneath databases and stateful workloads where downtime is measured in money and latency is measured in microseconds.

Our operator and CSI driver are the part of that product customers actually touch. The operator turns roughly twenty custom resources into a running, self-healing storage cluster — nodes, devices, pools, snapshots, backups, replication, migrations, upgrades. The CSI driver turns PVC requests into NVMe-oF volumes and keeps them attached through failovers, reboots and path changes.

We are a small company, which means the same person designs a CRD in the morning and explains to a customer's platform team in the afternoon why their cluster did what it did. Both halves belong to the role. We're looking for someone early in their career but already solid in Go, who wants to spend the next few years becoming genuinely excellent at Kubernetes controllers and storage.

?️ What you'll do

- Write and extend our controllers — reconcilers that never block, complex processes modelled as multistep operations in persisted phases. You start on well-scoped controllers and grow into the ones that move data.
- Work on the CSI driver, from provisioning and attachment down to NVMe-oF paths, multipath states and mounts — including the unglamorous half that matters most: self-healing after something crashed halfway through.
- Design and evolve the API. CRDs are user experience: validation and immutability where they belong, typed phases and conditions, conversions that keep a shipped field from breaking a customer.
- Reproduce and fix real cluster behaviour. A drain that stalls, a migration that loses writes,



a node that never gets re-probed. You read the logs, build the reproducer, and write the regression test that proves it.
- Write the failing test first. Every fix starts with a test describing the wanted behaviour, at the level that proves it — fake clients, a fake simplyblock, or a live cluster with real NVMe devices.
- Work directly with customers. Join a support call, read their logs with them, explain what their cluster is doing in terms a DBA or platform team can act on, and carry the outcome back into an issue, a test, or a fix.

? What you bring

- Go in production. One to three years is the shape we have in mind: interfaces, contexts, goroutines and errors hold no surprises, and you've shipped code that survived contact with users.
- Kubernetes fluency as a user, curiosity as a developer. PVCs, StorageClasses, DaemonSets, RBAC and lifecycle should be familiar. Having written a controller is a strong plus; wanting to is the minimum.
- Linux fundamentals. Block devices, filesystems, mounts, /sys and /dev, and enough systems instinct to be suspicious of the right things.
- Testing as a habit, not a chore. You'd rather spend an hour making a failure reproducible than an afternoon guessing.
- AI-assisted engineering, with judgment. We use Claude daily for code and log analysis, reproductions, refactors and documentation.



Use it to make yourself faster, and have opinions about where it quietly doesn't help.
- Composure in front of a customer. You listen before diagnosing, separate what's known from what's still a guess, and say you don't have an answer yet rather than promising a date.
- Excellent written and spoken communication with technical customers. You stay precise and calm when the other side is not.
- A proactive, high-responsiveness mindset. You chase the loose end nobody assigned to you, and you close the loop without being asked.
- Minimum 5 years of experience in customer-facing IT support roles; ideally working with distributed systems or Kubernetes
- Fluent in English

Preferred

- The CSI specification, or prior work on a storage or networking plugin
- NVMe, NVMe-oF, iSCSI, SPDK, or other userspace / kernel-bypass storage stacks
- kubebuilder, envtest, Ginkgo, or client-go beyond the basics
- Helm chart authoring, OLM bundles, or OpenShift
- gRPC, and reading an OpenAPI-generated client without flinching
- Python, JavaScript/TypeScript, C/C++ or Bash
- Prior customer-facing experience at an infrastructure or storage vendor
- Fluent English required; German or other European languages a plus

? What we offer

- Direct impact on a technically hard product, with a short path from your findings to shipped changes
- Work alongside the engineers who wrote the code. No escalation wall between you and them.
- The AI tooling budget to match a Claude-native company: Claude Max, Claude Code, and whatever else you can justify
- Competitive salary + equity
- A team that celebrates ambition and outstanding results. No participation trophies.

📌 Kubernetes Operator & CSI Engineer (Argentina)
🏢 simplyblock
📍 Argentina