Senior Systems Software Engineer, Compute Stack Acceleration

NVIDIA is growing a senior engineering team focused on making our compute software stack first-class on NVIDIA CPU platforms.

The team turns modern toolchains, build and code-health practices, performance-analysis workflows, and optimization techniques into repeatable improvements across real software components.

We are looking for an experienced systems software engineer who can lead cross-stack engineering efforts from an ambiguous adoption problem to a measurable outcome.

You will work closely with compiler, platform, performance, library, and application teams to validate new capabilities, resolve integration blockers, produce credible before-and-after evidence, and turn successful approaches into reusable engineering practices.

Depending on your background and project needs, your initial work may emphasize toolchain, build, and code-health adoption or profiling, optimization, and evidence-routing workflows.

The role is intentionally broad enough to evolve as the team identifies the highest-leverage opportunities across the stack.

• Lead toolchain, build, code-health, and performance-workflow adoption projects across large software components.
• Evaluate and deploy supported GCC and LLVM/Clang toolchains, compiler options, linkers, sysroots, and cross-compilation configurations.
• Integrate modern toolchains and workflows into complex build systems and continuous-integration environments.
• Establish useful Clang diagnostic builds and targeted sanitizer coverage in partnership with component owners. Evaluate techniques such as link-time optimization, profile-guided optimization, AutoFDO, and BOLT on representative software.
• Use profiling, PMU data, flamegraphs, and binary/source analysis to identify actionable performance and code-quality findings.
• Build automation, wrappers, validation scripts, dashboards, and migration helpers where they improve adoption and repeatability. Measure changes in runtime, code size, build time, launch latency, throughput, quality, or engineering velocity.
• Drive complex blockers to the appropriate compiler, runtime, library, infrastructure, or component owner.
• Document validated approaches as reusable playbooks for other engineering teams. Communicate technical results and tradeoffs clearly to engineers and leadership.