Performance Engineer (Kernel and Memory)

• Delivering this platform with high performance and efficiency depends on advanced Linux kernel and hardware capabilities, and that is where this team works
• The Performance Team owns overall system and application performance across that fleet: resource optimization, instrumentation, and the tooling that lets us see what the hardware is actually doing
• System memory sits alongside CPU cycles, network processing, and storage IOPS as a first-class resource, and how the kernel manages it is increasingly the thing that decides how hot we can safely run every machine we own
• We work upstream first
• Several of us maintain or regularly contribute to the Linux kernel, and our default is to fix problems in mainline rather than carry private patches
• This role is built on that principle
• Using memory efficiently is one of the highest-leverage problems we have
• At our scale, how the kernel’s memory-management subsystem behaves under pressure has direct consequences for both performance and cost
• As DRAM prices climb, getting more out of the memory we already own is a real lever on infrastructure spend
• Using custom low-overhead instrumentation on production servers, we have identified several scalability bottlenecks in the kernel memory layer that show up under real workloads
• We are looking for an expert to take on these problems, fix them against production workloads, and drive the changes upstream
• Solving them lifts efficiency and latency across the entire fleet rather than in one place
• Diagnose and fix performance and scalability bottlenecks in the Linux kernel memory-management subsystem, informed by real production telemetry at a scale few environments can offer
• Land your work upstream. We expect MM changes to go to mainline first and want you to keep doing the maintainership and review work that earns standing in the community
• Build and extend low-overhead kernel instrumentation (tracepoints, BPF, perf) so we keep measuring before we change, and prove the impact afterwards
• Partner with kernel, networking, storage, and product engineers across the company to turn observed system behavior into concrete MM improvements
• Set technical direction for how Cloudflare uses, tunes, and evolves kernel memory management as we push utilization higher

Benefits

• Competitive pay: We offer a competitive total rewards package, where every employee is an owner of our stock.
• Take-what-you-need vacation: We encourage employees to find a comfortable work-life balance by taking as many days off as they need while still being able to perform their jobs satisfactorily. (We really mean it!)
• Paid maternity & paternity leave: Our global parental leave policy allows up to sixteen paid weeks of bonding leave time for all qualifying new parents.
• Employee benefits: We offer a comprehensive benefits package including healthcare, life insurance, short- and long-term disability, pension plans in accordance with the market practice in our locations.
• Commuter program: We are flexible about where we work and value connecting at the office. Our commuter benefits program is in place to support team members’ commute to work via public transportation without the extra cost burden.
• Wellbeing: Global offerings that provide mental health, childcare and family forming support to our employees across our offices.- Deep expertise in the Linux kernel memory-management subsystem: the page allocator, virtual memory and memory mapping, reclaim, and the locking and concurrency that hold it together
• A track record of upstream Linux contribution, ideally including maintainership or sustained work in MM or an adjacent subsystem
• Clear communication with engineers across many teams, since the symptoms of an MM problem usually show up far from the MM code
• Strong systems programming in C, with the judgement to make high-risk changes to shared infrastructure safely
• Ability to dive deep into a complex codebase while keeping the fleet-wide picture in mind, and to own a problem and its outcome rather than a single patch
• Strong understanding of Linux kernel internals pertaining to memory management
• Experience troubleshooting and resolving performance issues in large-scale distributed systems
• Experience with large scale configuration/deployment management