Staff Software Engineer - Infrastructure Storage

Staff Software Engineer - Infrastructure Storage

Lambda is a leading cloud infrastructure provider focused on artificial intelligence workloads, serving tens of thousands of customers from independent researchers to enterprises and large-scale operators. The company's vision centers on democratizing access to high-performance compute, making advanced computational resources as accessible as traditional utilities. As AI and machine learning workloads continue to drive demand for specialized infrastructure, Lambda builds the systems that power cutting-edge research and production deployments at scale.

Lambda's Infrastructure Engineering team tackles the complex challenge of uniting storage, networking, and compute to create high-performance AI clusters. While raw GPU and CPU capacity matters, the real bottleneck in distributed AI systems lies in how quickly and reliably data moves to where computation happens. This team engineers the storage protocols, networking solutions, and virtualization layers that allow researchers and engineers to focus on their AI work rather than infrastructure management.

What you'll do

• Set technical direction for storage software architecture handling petabyte-scale deployments, author and review design documentation, mentor senior engineers, and lead cross-functional work spanning storage, networking, compute, and control plane teams
• Design, develop, and maintain high-performance storage systems across file protocols (NFS, SMB, Lustre), block protocols (NVMe-oF, iSCSI), and object storage (S3), including distributed orchestration and hardware integration
• Partner with storage, networking, control plane, Kubernetes, observability, and fleet engineering teams to deliver infrastructure initiatives, define storage performance targets (SLOs/SLIs), and ensure reliable deployment and operation
• Troubleshoot complex production issues spanning performance, protocol behavior, and hardware failures, owning the full lifecycle from requirements and design through deployment, monitoring, and maintenance
• Stay current with AI and HPC storage research, evaluate emerging protocols and hardware, and optimize storage solutions for AI-specific workloads including model checkpointing, dataset serving, and inference

What you'll bring

• Ten or more years in storage systems engineering, with at least five years in a technical lead or Staff-level individual contributor role
• Proven track record designing and operating multi-petabyte storage infrastructure in production data center or cloud environments, ideally with HPC, AI, or large-scale cloud storage background
• Strong systems programming proficiency in C, C++, Rust, or Go, with demonstrated ability to write high-performance, concurrent production code
• Deep hands-on experience with two or more storage protocols (S3, iSCSI, NVMe-oF, NFS, SMB, Lustre, or DAOS), including implementing or maintaining protocol implementations in production
• Experience profiling and optimizing storage performance for throughput, latency, and IOPS using tools like fio and elbencho
• Comfortable in physical data center environments, familiar with storage hardware, failure domains, rack-scale infrastructure, and storage observability tools (Prometheus, Grafana, log aggregation, tracing)
• Working knowledge of NVMe, NVMe-oF, RDMA (RoCE or InfiniBand), and DPU technology

Nice to have

• Experience with NVIDIA BlueField DPUs or SuperNICs for accelerated storage data paths and GPUDirect Storage
• Production experience with enterprise or HPC storage platforms including Vast Data, Weka, NetApp, or Lustre
• Large-scale Ceph deployment and operations experience (100PB+) in HPC or AI environments
• Exposure to emerging storage technologies such as CXL memory pooling, computational storage, or ZNS SSDs
• Contributions to or maintenance of open-source storage projects (Ceph, DAOS, Lustre, MinIO)

What they offer

• Full-time position based in the San Francisco office on Fremont Street with hybrid work arrangement
• Core office presence required four days per week, with Tuesday designated as remote work day