Description
We're looking for a Senior Engineer for CoreWeave's Benchmarking & Performance team. You will have an integral part in our planet-scale performance data warehouse: Ingesting, storing, transforming and analysing performance events in all the data centres across our global infrastructure. You will also aid us in achieving industry-leading end-to-end performance benchmarking publications such as MLPerf.
You will be an owner who leads designs, raises engineering standards, and delivers measurable improvements to latency, throughput, and reliability across multiple services. You'll partner with product, orchestration, and hardware teams to evolve our Kubernetes-native platform and meet strict P99 SLAs at scale.
Key responsibilities include:
- Developing and enhancing Kubernetes-native benchmarking services that measure latency, throughput, jitter, and cost-per-request across CoreWeave's compute stack.
- Contributing to implementing and maintaining benchmarking workflows for end-to-end MLPerf Training and Inference runs, including workload setup, cluster configuration, and result validation.
- Participating in design discussions and contributing to architecture decisions within the team.
- Breaking down engineering tasks into clear milestones and delivering reliable, high-quality code.
- Collaborating with teammates to maintain reproducible, well-documented benchmarking processes.
- Providing constructive code reviews and sharing best practices with peers.
- Mentoring junior engineers; reviewing cross-team designs and elevating coding/testing standards.
Requirements include:
- 3–5 years of experience building distributed systems, high-performance computing components, or cloud services.
- Strong programming skills in Python or Go (C++ a plus) with understanding of networked systems and performance fundamentals.
- Hands-on experience with Kubernetes in production environments plus familiarity with CI/CD and observability tools (e.g., Prometheus, Grafana, OpenTelemetry).
- Exposure to performance-critical GPU systems (CUDA, NCCL, NVLink/PCIe, memory bandwidth) or model-serving stacks (llm-d, vLLM, TensorRT-LLM, Megatron-LM).
- Effective communicator comfortable working cross-functionally.
Nice to have:
- Experience with time-series databases, LSM-based storage engines, or custom data pipelines.
- Familiarity with MLPerf or other large-scale benchmarking frameworks.
- Contributions to OSS projects such as llm-d, vLLM or PyTorch.
- Exposure to benchmarking GPU clusters or multi-region environments.
- Background working with CUDA kernels, NCCL/SHARP, RDMA/NUMA, or GPU interconnect topologies.