Senior Engineer, Storage (File & Block)

CoreWeave•San Francisco, CA

47d•Onsite

About The Position

We're looking for a Senior Storage Engineer, File & Block to help build and operate the file and block storage services at the heart of CoreWeave's Storage team — the high-performance, multi-tenant storage that our largest AI training and inference workloads and internal stateful services run on. You'll help turn CoreWeave's storage from a follower of GPU growth into a multiplier of it: owning the data path and control plane for both file and block, running them on our own hardware fleet, and shipping the customer-facing capabilities that unblock petabyte-scale deals. You'll work closely with compute, platform, and infrastructure teams to make storage fast, reliable, and effortless to consume at scale.

Requirements

Bachelor's or Master's degree in Computer Science, Engineering, or a related field.
6–10 years building storage systems, distributed systems, or infrastructure services.
Strong hands-on experience with distributed or networked file systems and/or block storage in production — e.g., NFS/POSIX file semantics at scale, and/or block volume services (durability, snapshots, replication).
Depth in the storage data path: IO performance, caching, small-file and metadata-heavy workloads, crash consistency, and resilience under load.
Experience with multi-tenant services or storage control planes — provisioning, tenancy/quota, and data-protection features (snapshots, encryption, key management).
Proficiency in a systems/back-end language — Go strongly preferred (C or Rust a plus).
Experience building on Kubernetes — controllers/operators, CRDs, and CSI (dynamic provisioning, resize, snapshots, RWO volumes).
Familiarity with distributed databases (e.g., CockroachDB), workflow orchestration (e.g., Temporal), and gRPC/Protobuf service design.
Ideally, experience with distributed or parallel storage stacks such as Ceph/RBD, Lustre, GPFS/Spectrum Scale, BeeGFS, WEKA, VAST, or DAOS.
Familiarity with storage observability tools and telemetry pipelines (e.g., ClickHouse, Prometheus, Grafana).
Bonus: exposure to high-performance data-path technologies (RDMA, GPUDirect Storage, RoCE, InfiniBand, SPDK).
Strong debugging and problem-solving skills in distributed, high-performance environments.
Clear communicator, able to work collaboratively across teams and share technical insights effectively.

Responsibilities

Design, build, and operate highly scalable, multi-tenant file and block storage — from the data path (NFS for file; durable, high-performance block volumes) to the control plane that governs tenancy, provisioning, and data protection — running natively on CoreWeave's storage fleet and delivered through Kubernetes.
Own the file system data path: NFS hot-path reliability, IO resilience, and performance — including the small-file / IOPS-heavy workloads that AI pipelines depend on, and the throughput that keeps GPUs fed.
Build a native block storage service: durable, high-performance block volumes for customer workloads and internal stateful services — the third leg of the file/object/block triad, running on our own data path and shared storage hardware.
Deliver both file and block as first-class Kubernetes citizens: design and ship a backend-neutral CSI driver with dynamic provisioning, resize, snapshots, and RWO block volumes, backed by a public provisioning SLO — reusing shared control-plane and CSI patterns across file and block rather than rebuilding per service.
Build the control plane: multi-tenant isolation, quota, QoS, snapshots, key management (BYOK), audit, and lifecycle policy — the capabilities that unblock regulated and enterprise customers (e.g., encryption in transit, HIPAA controls).
Improve the reliability, durability, and observability of the storage stack; partner with operations to monitor, analyze, and optimize using telemetry, metrics, and dashboards to improve performance, latency, and resilience.
Work cross-functionally with platform, product, and infrastructure teams to deliver seamless storage across the stack, including tiering cold data to lower-cost object storage while preserving access.
Share your knowledge and mentor other engineers on best practices in building distributed, high-performance systems.