Staff Software Engineer - Linux/Kernel at datadog

The role

The eBPF APM team builds Datadog’s zero-instrumentation observability platform, enabling automatic service discovery, Layer 7 protocol classification, and traffic decoding. This platform provides high-fidelity RED metrics from both plaintext and TLS-encrypted traffic without requiring customer code changes. The work spans kernel-space eBPF, user-space Go services, and large-scale distributed systems, operating reliably across diverse Linux environments.

What you'll do

• Own major technical areas of the zero-instrumentation APM system, from design through long-term evolution and operational maturity.
• Define and drive architecture for kernel-level traffic capture, L7 protocol decoding, and metric extraction using eBPF and Go.
• Lead high-impact initiatives addressing protocol parsing, TLS visibility, kernel compatibility, and performance at scale.
• Set technical direction and standards for reliability, performance, and maintainability across the team.
• Partner cross-functionally with Agent, Tracing, Security, Runtime, and Product teams to align on system design and roadmap priorities.

What you'll bring

• You have deep experience in backend or systems engineering, with strong proficiency in Go and/or C/C++.
• You are comfortable operating close to the Linux kernel, with experience in eBPF, networking, observability, or similarly low-level systems.
• You consistently think at a system and organizational scale, making thoughtful tradeoffs between performance, correctness, velocity, and long-term sustainability.
• You have built, evolved, or operated large-scale production systems in complex and heterogeneous environments.
• You demonstrate technical leadership without relying on authority, influencing architecture and raising the technical bar through collaboration.

Your impact

At Datadog, Staff Engineers are expected to deliver impact through technical leadership, not just individual contribution. This role offers the opportunity to shape a foundational observability product used at massive scale.