FluidstackSan Francsisco, CABuild and own the observability platform, control plane APIs, and fleet state management for a hyperscale GPU infrastructure powering AI compute at 10-100s of GW scale. Requires production service ownership at scale, comfort with AI coding tools, and on-call incident response.
Own the observability platform for the GPU fleet. Build and operate data pipelines, decoration and correlation engines, and healthcheck frameworks that make the fleet legible from site-level down to individual device and link.
Define and build the API surface for infrastructure, including contracts between production infrastructure and all dependent tools. All other teams use this tooling to manage and operate the hyperscale fleet.
Build the production control plane, including unified machine management, actual state inspection, distributed command execution, and the Kubernetes-based infrastructure that underpins it.
Own fleet state as the source of truth, including SLOs, site lifecycle state, and integrations with internal infrastructure management and customer-facing operations platforms. Ensure the system's view of itself always matches reality.
Land new hardware into the platform cleanly, handling ZTP, DHCP, DNS, artifacts, and integration for every new XPU generation and site.
Bonus skills: Distributed systems and data pipeline engineering. Time-series observability stacks (Prometheus, Thanos, VictoriaMetrics). API design and versioning at scale. Workflow and orchestration engines (Temporal, Cadence). BMC/Redfish or hardware telemetry. Go, Python, and Postgres.
Own end-to-end health, reliability, and automation for a massive GPU compute fleet at Fluidstack. Build metrics, repair pipelines, GPU qualification platforms, and low-level BMC/Redfish tooling while driving incidents and using AI coding tools daily. Hardware intuition and comfort with ambiguity required.
Own observability, API surface, and control plane for a hyperscale AI compute fleet. Build production-grade data pipelines, stateful APIs, and Kubernetes infrastructure that other teams depend on.
Own end-to-end health, repair automation, and qualification of a hyperscale GPU/TPU compute fleet. Build metrics pipelines, firmware tooling, and self-healing repair workflows across Kubernetes and bare metal.
Builds and scales core infrastructure including ML training/serving, Kubernetes clusters, and low-latency voice/audio pipelines. Requires 3+ years in infrastructure/ML systems, hands-on reliability engineering, and Kubernetes expertise.
Builds and operates reliable, scalable AI infrastructure including observability, SLOs, incident response, automation, and performance tuning for ultra-low-latency serverless compute. Requires 3+ years SRE/DevOps experience with cloud, Kubernetes, programming (Go/Rust/Python), and observability tools.