> Markdown version of [/jobs/ext/4628263-software-engineer-cloud-infrastructure](https://www.wearedevelopers.com/jobs/ext/4628263-software-engineer-cloud-infrastructure). Every page supports `.md` or `Accept: text/markdown`. Links point to the HTML versions so they work for humans too. Agent guide: [/agents.md](https://www.wearedevelopers.com/agents.md).

---

# Software Engineer, Cloud Infrastructure

- **Company:** Fluidstack Ltd
- **Location:** New York, NY, United States
- **Salary:** $175,000 - $300,000
- **Contract:** Permanent contract
- **Skills:** Application Programming Interfaces (APIs), Cloud Computing, Dynamic Host Configuration Protocol, Domain Name System (DNS), Infrastructure as a Service (IaaS), Graphics Processing Unit (GPU), Kubernetes, Data Pipelines
- **Published:** November 7, 2026
- **Apply:** https://www.juju.com/job/28_25303_82d5a0b0a559fe15630c74100f9fa1f2

## About the Role

We value candidates who treat toil as a bug, design APIs that age well, move toward ambiguity, learn at a steep slope, carry a pager without flinching, are fluent with AI tooling, and have shipped production services that other teams depend on at scale.

## Description

* Make tens of thousands of GPUs legible in real time: build the observability platform that turns raw telemetry into signal, from site-level health down to individual device and link. At 10 GW scale, you cannot operate what you cannot see.
 * Build the control plane every team at Fluidstack depends on: replace one-off tooling with a stable, versioned API surface that covers unified machine management, actual state inspection, and distributed command execution. One interface for the whole company, not a hundred scripts.
 * Make the system's view of itself always match reality: integrate fleet state as a machine-readable source of truth across provisioning, operations, and customer-facing platforms, so every new site and GPU generation lands cleanly from day zero.

Role Scope

 * Own the observability platform. Build and operate the data pipelines, decoration and correlation engine, and healthcheck framework that make the fleet legible - from site down to device and link. No other team should need to scrape production directly to answer a question.
 * Define and build the API surface for infrastructure. Design the contracts between production infrastructure and every tool that touches it. All other teams at Fluidstack use your tooling to manage and operate our hyperscale fleet.
 * Build the production control plane. Unified machine management, actual state inspection, distributed command execution - and the Kubernetes-based infrastructure that underpins it all.
 * Own fleet state as source of truth. SLOs, site lifecycle state, and integration with internal infrastructure management and customer-facing operations platforms. What the system says about itself should match reality, and you're accountable when it doesn't.
 * Land new hardware into the platform cleanly. ZTP, DHCP, DNS, artifacts - every new XPU generation and site integration goes through IaaS before production.