Staff Software Engineer, Observability & Profiling
Core
Design and build scalable telemetry ingest and storage pipelines, observability solutions, and instrumentation libraries for Anthropic's massive GPU, TPU, and Trainium clusters to ensure operational reliability.
Role type
Staff Software Engineer (Infrastructure/Observability)
Builds
High-throughput telemetry pipelines, fleet-wide continuous profiling, eBPF-based tracing, and AI-assisted diagnostic tooling
Domain
Cloud Infrastructure, Observability, High-Performance Computing (GPU/TPU/Trainium)
Deliverable
production ML models | infrastructure
Required skills
Large-scale observability infrastructure design, end-to-end telemetry signal management, high-throughput pipeline architecture, kernel/network stack debugging, eBPF development, continuous profiling at fleet scale, accelerator workload instrumentation, OpenTelemetry expertise
Preferred skills
eBPF-based observability in production, fleet-scale continuous profiling management, kernel/syscall-level debugging, accelerator profiling, high-cardinality metrics systems, AI/LLM application to operational workflows
Technologies
eBPF, OpenTelemetry, GPU/TPU/Trainium clusters, kernel stack, network stack
Responsibilities
Design scalable telemetry pipelines for metrics, logs, traces, and error data; Build observability solutions for deep system visibility; Own and evolve core observability platforms; Build instrumentation libraries and SDKs; Reduce MTTR via cross-signal correlation and unified query interfaces; Drive fleet-wide efficiency through continuous profiling insights
Seniority
Staff, hands-on IC with strategic impact
