Staff Software Engineer - Managed Kubernetes

Lambda•San Francisco, CA

12d•$314,000 - $465,000•Hybrid

About The Position

Lambda is building the AI Cloud of the future and is seeking a Staff Engineer to help develop its Managed Kubernetes platform, designed for AI workloads on bare metal. This is a foundational technical leadership role focused on shaping the infrastructure for AI training and inference at scale. The engineer will collaborate on the technical vision for managed orchestration services, including Managed Kubernetes, Managed Slurm on Kubernetes, and higher-level platform services for inference and AIOps. The role involves working at the intersection of distributed systems, GPU-accelerated computing, and Cloud Native infrastructure to build reliable, performant, and simple systems. It requires a holistic understanding of infrastructure domains (compute, network, storage, security) and the ability to design solutions across them, with close collaboration with NVIDIA's open-source ecosystem and internal teams.

Requirements

10+ years of experience in software engineering, platform engineering, or SRE, with at least 5 years focused on Kubernetes at scale.
Expert-level understanding of Kubernetes internals: API machinery, controllers, schedulers, operators, CRDs, CSI, CNI, and the extension patterns that make Kubernetes powerful.
Holistic infrastructure expertise: you've synthesized knowledge across compute, networking, storage, and security, not just Kubernetes in isolation. You can build solutions that span the full stack.
Strong software engineering skills in Go (required) and Python; you write production-quality code, not just scripts.
Deep experience with GPU orchestration in Kubernetes: NVIDIA GPU Operator, device plugins, DCGM, MIG, time-slicing, and GPU-aware scheduling.
Proven track record of technical leadership: driving design decisions across teams, mentoring engineers, and influencing infrastructure direction beyond your immediate scope.
Deep experience designing and operating managed services or multi-tenant platforms. You understand what it takes to run infrastructure for external customers.
Strong understanding of distributed systems principles: consensus, fault tolerance, consistency models, and graceful degradation.
Experience with observability at scale: Prometheus, Grafana, distributed tracing, and building actionable alerting systems.
Solid knowledge of Linux systems and networking (L2-L7), including high-performance networking concepts (RDMA, InfiniBand, RoCE).
Experience with infrastructure-as-code and GitOps workflows.

Nice To Haves

Familiarity with NVIDIA Network Operator and GPUDirect is strongly preferred.
Experience building and operating managed Kubernetes services (GKE, EKS, AKS, or similar) or working on Kubernetes control plane components.
Hands-on experience with NVIDIA's open-source ecosystem beyond GPU Operator: Network Operator, NCCL tuning, Topograph, AICR, or similar emerging projects.
Familiarity with HPC and traditional job schedulers (Slurm) and Kubernetes-native batch scheduling (KAI, Volcano, Kueue).
Background in confidential computing.
Experience migrating customers or workloads from legacy/bespoke infrastructure to standardized platforms.
Contributions to CNCF projects, Kubernetes SIGs, or NVIDIA open-source projects.
Familiarity with security and compliance in multi-tenant environments: RBAC, Pod Security Standards, network policies, workload isolation.
Background in ML infrastructure: training clusters, inference serving, simulation.

Responsibilities

Drive technical vision for Lambda's Managed Kubernetes bare-metal platform, including control plane scalability, multi-tenancy, cluster lifecycle management, and high availability.
Integrate and extend NVIDIA's open-source ecosystem: GPU Operator, Network Operator, DCGM, NCCL, and emerging projects like AICR and Topograph for topology-aware scheduling and placement.
Design GPU-aware orchestration systems.
Lead development of services that power our managed services.
Inform on and help with networking solutions for AI workloads: CNI integration (Cilium, Multus), high-performance fabrics (InfiniBand, RoCE), RDMA, and GPUDirect. You will work closely with our Network team to define and drive requirements.
Inform and help with storage architecture requirements for AI workloads. You will partner with Storage teams on what managed K8s, Slurm, and future services need.
Build the foundation for Managed Slurm on Kubernetes, enabling traditional HPC workloads to run seamlessly alongside Kubernetes workload.
Design higher-level platform services for inference, including model serving infrastructure, autoscaling based on inference load, and multi-model deployment patterns.
Design self-healing systems and automation for incident response, root cause analysis, and platform resilience.
Lead chaos engineering efforts to validate system behavior under failure conditions at scale.
Establish operational excellence for a managed service: upgrade automation, security patching, and zero-downtime maintenance.
Serve as the technical bridge between Orchestration and other infrastructure teams (Network, Storage, Security), translating platform requirements into actionable specifications.
Drive infrastructure-wide decisions that enable successful managed services. You’re someone who understands what's needed end-to-end, not just at the Kubernetes layer.
Provide input on bare-metal provisioning, network topology, and storage systems to ensure they meet the needs of managed the services being built by the Orchestration organization.
Champion consistency and standardization across Lambda's infrastructure stack.
Work directly with customers and internal teams to understand existing deployments and chart a path to the managed platform.
Set technical direction for Kubernetes services across the Orchestration team, influencing roadmap and prioritization.
Drive reviews and design sessions, ensuring we build systems that are scalable, maintainable, and aligned with customer needs.
Mentor and grow engineers, establishing best practices for Kubernetes development, distributed systems, and Cloud Native engineering.
Collaborate cross-functionally with Network, Storage, Security, and Customer Success teams.
Engage with NVIDIA and the open-source community to stay current on GPU orchestration technologies and contribute back where appropriate.
Represent Lambda externally through technical blog posts, conference talks, and strategic customer engagements.
Shape our AIOps vision: design intelligent systems for automated capacity planning, anomaly detection, and predictive maintenance of cloud infrastructure.