Staff Software Engineer, Observability & Profiling

About the role

Anthropic is looking for a Staff Software Engineer to join its Observability team within the Infrastructure organization. The team builds and operates the monitoring and telemetry platform that powers research and product systems across massive GPU, TPU and Trainium clusters.

Key responsibilities

• Design and implement scalable telemetry ingest and storage pipelines for metrics, logs, traces, and error data across multi‑cluster infrastructure.
• Develop observability solutions that provide low‑overhead, deep visibility into system behavior fleet‑wide.
• Own and evolve core observability platforms, driving migrations, architectural improvements, and cost reductions.
• Build instrumentation libraries, SDKs, and eBPF‑based auto‑instrumentation to emit high‑quality telemetry.
• Reduce mean time to detection and resolution by creating cross‑signal correlation, unified query interfaces, and AI‑assisted diagnostic tools.

Required profile

• Hands‑on experience building and operating large‑scale observability or monitoring infrastructure.
• Deep understanding of end‑to‑end observability signals, from instrumentation through ingest to analysis.
• Experience with high‑throughput telemetry pipelines and the trade‑offs of collecting, storing, and querying data at scale.
• Comfort digging below the application layer into the kernel, network stack, or hardware.

Required skills

• eBPF‑based tracing, profiling, and network visibility.
• OpenTelemetry instrumentation and collector pipelines.
• Continuous profiling at fleet scale.
• High‑cardinality metrics systems and large‑scale telemetry storage.
• Kernel‑level debugging and performance engineering.
• Profiling and instrumentation of accelerator workloads.