C++ Engineer — Pushing the Speed of Light | Ultra-Low Latency Trading Systems

We are partnering with one of the world’s most technically ambitious proprietary trading firms - a group rebuilding their entire trading platform from the metal up to operate at the physical limits of modern hardware. This isn’t an incremental improvement. It’s a total re-architecture of the fastest system on the planet, where every microsecond is contested ground and every cache miss is a bug. Their engineers operate where nanoseconds decide P&L — measured, profiled, and deployed in live markets where performance is the edge. They’re now seeking an elite C++ Engineer capable of designing and optimising the core of a real-time execution platform — a system that ingests millions of market events per second and reacts deterministically, faster than anyone else on Earth. The successful engineer will: Architect zero-GC, lock-free pipelines built around ring buffers and cache-aligned data structures. Develop custom kernel-bypass network stacks using DPDK, RDMA, and Solarflare Onload , tuned to single-digit microsecond latency. Engineer branch-prediction-aware order handlers and SIMD-vectorized pricing logic in AVX-512 . Deliver next-tick telemetry with nanosecond-precision timestamps and cross-core synchronization. Collaborate with FPGA specialists to merge hardware precision with software agility. The Toolkit Modern C++20/23 , template metaprogramming, constexpr, inline assembly when necessary. Profiling and optimization using perf , VTune , bcc , and FlameGraphs . Deep knowledge of NUMA-aware design , memory fences, and lock-free concurrency. Expertise in custom allocator design , branchless algorithms , and profile-guided optimization . A habit of benchmarking rather than assuming — data, not theory. Ideal Background Proven experience building ultra-low-latency systems in trading, gaming, or networking. Deep understanding of CPU architecture , from cache hierarchies to speculative execution. The mindset of someone who thinks in nanoseconds and measures in CPU cycles . A record of winning battles with compilers, kernels, and performance bottlenecks. This firm operates on a flat structure — no committees, no bureaucracy, no excuses. Engineering, hardware, and trading sit shoulder-to-shoulder. Code that’s 10ns faster doesn’t just run better — it changes the business. If you believe latency is the final frontier, and profiling is the only truth, this is the environment you’ve been building toward. #J-18808-Ljbffr Mondrian Alpha