CoreWeave
Staff Software Engineer, Inference
$188K–$275K
Sunnyvale, CA / Bellevue, WA +2 · Staff · Full-time · Posted May 7, 2026
- Location
- Sunnyvale, CA / Bellevue, WA
Bellevue, WA
Sunnyvale, California, United States - Experience
- 8–12+ years building and operating large-scale distributed systems or cloud platforms; experience leading cross-team technical initiatives is also required.
- Visa sponsorship
- Requires eligibility to access export-controlled information: U.S. person status, eligibility to access without export authorization, or eligibility and reasonable likelihood of obtaining required authorization. CoreWeave may decline to pursue export licensing.
About the job
Provides technical leadership for CoreWeave’s Kubernetes-native inference platform, improving architecture, performance, and reliability across services and teams.
What you'll do
- Lead cross-team design initiatives and influence engineering direction across the organization.
- Optimize inference latency, throughput, and GPU utilization, including scheduling, batching, and memory use.
- Improve reliability and performance of distributed systems at scale using metrics-driven approaches.
What they ask for
- Strong programming skills in Go, Python, or C++.
- Deep production-scale Kubernetes expertise, including orchestration, scheduling, and service design.
- Experience with distributed systems, networking, performance optimization, and low-latency, high-throughput systems with strict P95/P99 latency requirements.
- Hands-on inference-systems experience, including batching or micro-batching, caching, and memory optimization.
Nice to have
- Experience with inference frameworks such as vLLM, Triton, TensorRT-LLM, Ray Serve, or TorchServe.
- Experience with GPU performance optimization or large-scale AI/ML infrastructure.
Pay
$188,000 to $275,000 USD base salary
Tools
Go, Python, C++, Kubernetes, vLLM, Triton, TensorRT-LLM, Ray Serve
AI focus
Builds and optimizes AI inference infrastructure, including inference performance, GPU utilization, batching, memory optimization, and streaming workloads.
Model-written summary · the original posting is authoritative