Systems Engineer - Network Software Performance
- Location
- Hybrid
- Salary
- $168K–$210K / yr
- Posted
- November 20, 2026
- Source
- Greenhouse
Job description
About Us
Locations: Austin TX, Champaign IL, Lisbon Portugal, London England, Atlanta GA, Washington DC,
About the team
Cloudflare is redefining the way networks are run at scale, and some of the largest networks in the world now run their traffic through us. Cloudflare's network runs in hundreds of cities on general-purpose Linux servers, not dedicated routers, so the routers, firewalls, load balancers, and proxies that carry our customers' traffic are software that we write, ship, and operate ourselves.
Our Packet Processing team builds and operates that software: high-performance code that processes customer traffic at the packet level as it moves through Cloudflare's network, in every Cloudflare data center. That includes the L4 load balancer at the front door of every server, a per-customer routing and forwarding data plane, a stateful network firewall, and a TCP/UDP reverse proxy (the engines behind products like Magic Transit, Cloudflare WAN, Cloudflare Network Firewall, and Spectrum).
We are a team of engineers who obsess over performance, correctness, and reliability at massive scale. We own what we build in production, a lot of our work happens close to the kernel and the NIC.
About the role
For us, performance is a product feature. Customers notice single-flow throughput, packet loss, and latency right away, and every CPU cycle we spend on a packet is multiplied across every server in our network. We're looking for engineers who love finding out where the time goes, and then making it go away.
Some of the things you'd work on
- Pushing single large flows (e.g., one tunnel carrying a customer's whole data center) to multiple Gbps and beyond, and making that performance consistent
- Finding out exactly where packets get dropped and why: NIC queues, interrupts and softirq, XDP, veth and namespaces, userspace rings, and everything in between
- CPU placement: NUMA, RSS, core isolation, and steering traffic to the right cores on servers that also run many other services
- Cutting the CPU cost per packet in hot paths written in Rust, C, and Go: batching, cache-friendly data structures, fewer copies, fewer locks
- Measuring all of it: profiling with perf, eBPF tracing, and flame graphs; continuous profiling across the fleet; benchmarks in the lab and careful experiments in production
- Building the benchmark and test tooling that keeps performance regressions from shipping
- Helping shape a unified data plane platform so that it's fast from the start
We write mostly Go and Rust (plus C for eBPF) on Linux, and the work ranges from NIC configuration and the kernel network stack to userspace packet pipelines.
You'll build, deploy, and operate these systems on top of one of the largest edge networks in the world. You'll partner with some of the best engineers and product managers, on our team and across Cloudflare, to solve big problems that have a meaningful impact in the lives of our customers and users.
We're looking for curious, empathetic engineers who are excited to join us in changing the way networking is done!
Examples of desirable skills, knowledge, and experience
- Systems programming experience in Rust, Go, C, or C++
- Performance analysis on Linux: perf, eBPF-based tracing (e.g., bpftrace, BCC), flame graphs, hardware counters
- A working model of modern CPUs and memory: caches, NUMA, branch prediction, SIMD, and how they show up in a profile
- Knowledge of how packets move through Linux: NIC queues, RSS, NAPI, GRO/GSO, softirq, and the socket layer
- Experience making high-throughput or low-latency systems faster, and proving it with good measurements
- A solid understanding of L3 and L4 networking: IP, TCP, UDP, and tunneling
- Building and operating large-scale, reliable, distributed systems
- Operating what you build: on-call, incident response, and root cause analysis
- Reading unfamiliar code and safely evolving systems that have been in production for years
- Owning work end to end, from design through deployment, and working across teams to land shared solutions
- Clear written and verbal communication; we design in written specs and review each other's work
Bonus points
- Tuning AF_XDP, DPDK, or other kernel-bypass systems
- Low-latency backgrounds such as trading systems, HPC, games, or storage
- Go runtime or Rust async performance work
- NIC offloads and features (e.g., checksum, TSO/LRO, flow steering, SR-IOV)
- Contributions to the Linux kernel or open-source performance tooling
- Compensation may be adjusted depending on work location.
- For Champaign IL based hires: Estimated annual salary of $168,000 - $210,000.
- For Washington, DC based hires: Estimated annual salary of $185,000 - $230,000.
- For Seattle, WA based hires: Estimated annual salary of $185,000 - $230,000.
- For Portugal based hires: Estimated annual salary is between €66,000 - 83,000.