About the roleYou will own systems across the lifecycle of our accelerator clusters-from bringing capacity online and upgrading fleets to detecting failures, recovering safely, and retiring capacity. Your work will determine how quickly researchers can start experiments, how efficiently expensive hardware is used, and how reliably long-running workloads complete.
This is a hands-on infrastructure role spanning cloud and datacenter environments, cluster control planes, networking, storage, security, observability, and automation. You will partner with hardware and cloud providers and our Training, RL Systems, Post-Training, Inference, and Security teams to turn heterogeneous compute into a dependable platform. Depending on experience, you may lead multi-quarter initiatives and help set technical direction.
What you'll do- Design, build, and operate control-plane services and infrastructure-as-code for provisioning, configuration, validation, upgrades, expansion, draining, recovery, and decommissioning; make every change repeatable, auditable, and safe to roll back.
- Bring new accelerator capacity online on schedule by coordinating dependencies across cloud providers, datacenter and hardware partners, networking, storage, security, and internal compute consumers.
- Build high-bandwidth, topology-aware connectivity within and across clusters; diagnose performance and reliability issues spanning hosts, switches, routing, transport, collective communication, and workload placement.
- Make clusters secure by default through identity and access controls, network policy, workload isolation, host and container hardening, secrets management, and trusted software and image supply chains.
- Improve fleet scalability, consistency, and fault tolerance by defining health signals, automating remediation, reducing configuration drift, and designing for partial failure.
- Establish service-level objectives and observability for cluster readiness, provisioning time, usable capacity, job-start latency, infrastructure-caused failures, utilization, and recovery time.
- Lead incident response and blameless postmortems for cluster failures; turn recurring operational pain into automation, safer defaults, and simpler system boundaries.
- Work directly with Training, RL Systems, Post-Training, and Inference engineers to debug cross-layer failures and shape a long-term compute, data, networking, and capacity roadmap.
You may be a good fit if you have- Deep expertise in distributed systems, reliability, and cloud platforms (e.g., Kubernetes, IaC, AWS/GCP/Azure).
- Strong programming ability in Python, Go, Rust, or another language suited to reliable infrastructure services and automation.
- Hands-on experience with Linux, containers, Kubernetes or another cluster scheduler, infrastructure-as-code, and at least one major cloud platform or substantial bare-metal environment.
- A practical understanding of networking, storage, identity, observability, and reliability, with the ability to trace a failure across multiple layers of a complex system.
- Experience designing systems for safe rollout, fault isolation, idempotency, capacity growth, and recovery from partial or large-scale failures.
- High ownership and clear communication, including comfort coordinating multi-team projects and participating in a healthy on-call rotation.
Strong pluses- Experience operating large GPU or accelerator fleets for distributed model training, inference, scientific computing, or another communication-intensive workload.
- Depth in Kubernetes internals, custom controllers or operators, device plugins, cluster autoscaling, scheduler extensions, Slurm, or comparable orchestration systems.
- Experience with high-performance networking such as RDMA, InfiniBand, RoCE, BGP, cloud interconnects, multi-NIC hosts, CNI or eBPF networking, or topology-aware placement.
- Experience with Terraform, workflow orchestration, and automated qualification of hosts, drivers, firmware, networks, and new hardware.
- Knowledge of GPU systems, NCCL, NVLink or NVSwitch, and the failure modes of large distributed jobs.
How we work- Mission first. We choose work for its impact on the mission and take responsibility for the outcome, not just our assigned tasks.
- High agency. We identify what is missing, form a plan, and move without waiting for perfect clarity.
- Speed with rigor. We ship, measure, and iterate quickly while protecting correctness, safety, and reliability.
- Flexible scope. We cross team and technical boundaries when that is the fastest way to solve the real problem.
- Low ego, high standards. We give direct feedback, change our minds when the evidence changes, and help the whole team win.
- Continuous learning. The stack changes quickly; we are willing to learn unfamiliar systems, methods, and domains as the work demands.
Location, visa sponsorship & benefits- Location-based hybrid policy. This is a location-based hybrid role. We currently expect all staff to work from one of our offices at least three days per week. Exact office options will be confirmed during the recruiting process.
- Visa sponsorship. We do sponsor visas. However, we cannot successfully sponsor a visa for every role and every candidate. If we make you an offer, we will make every reasonable effort to secure the necessary visa, and we retain immigration counsel to support the process.
- Meals and office benefits. We provide complimentary lunch and dinner in our offices, along with snacks and beverages.
A note on qualifications. We care more about exceptional evidence than a perfect keyword match. If the work excites you and you can show unusual strength, learning speed, or ownership, we encourage you to apply even if your background does not match every preferred qualification.
