Inside the Stack (CUDA / eBPF / Tokio)
CoreWeave runs Rust where garbage pauses and unpredictable memory growth are unacceptable. At 32,000-GPU training scale, a single degraded optical transceiver or silent XID GPU error stalls an entire synchronous training run. Our Rust node daemons monitor NVLink/InfiniBand counters via eBPF and NVML in real time.
Core Challenges on Your Plate
- Develop low-overhead Rust telemetry agents, automated topology-aware slurm/Kubernetes job placement plugins, and RDMA congestion diagnostics.
- Scale internal components leveraging CUDA, eBPF, Tokio, Tonic, no_std while keeping binary compile times and memory RSS tight.
- Instrument p99.9 latency metrics and lead root-cause incident analyses across distributed clusters.
Requirements & Verified Salary Band
- Deep familiarity with idiomatic Rust, async execution models, and systems debugging tools (`gdb`, `perf`, `ebpf`).
- Prior ownership of production storage engines, compilers, network proxies, or cryptographic runtimes.
- Base salary: $215,000 to $295,000 USD + equity. Location: US Remote / East Coast.
