Global Banking & Markets, Senior Low-Level Systems Engineer (Linux Kernel and Low Latency), Vice President , Hong Kong

Location: Hong Kong

Overview

We are seeking an experienced developer passionate about low-level coding to design and optimize high-performance, mission-critical systems on Linux. You will work close to the metal across kernel and user space, focusing on performance, reliability, and deterministic latency. Experience in financial markets is a strong plus, but not required.

What You Will Do

• Design, build, and optimize high-performance services in: C, C++, and Java.
• Engineer performance-critical components across user space and kernel interfaces, emphasizing memory, scheduling, I/O, and networking paths.
• Develop and tune networking pipelines, including multicast and unicast UDP and TCP, with careful socket, buffer, and NIC configuration.
• Implement kernel-bypass or fast-path networking where appropriate (for example, DPDK, netmap, XDP), including queue, NIC offload, and CPU affinity strategies.
• Contribute to or interact with Linux kernel subsystems: memory management, scheduler, device drivers, and filesystems (VFS).
• Optimize systems for latency and determinism: CPU isolation, thread pinning, NUMA locality, cache awareness, lock contention reduction, and memory allocator tuning.
• Apply distributed systems patterns such as sequencer (virtual synchrony) for total ordering and consistency where required.
• Profile, measure, and debug using tools such as perf, ftrace, eBPF, perf_events, tcpdump, and flame graphs to find and eliminate bottlenecks.
• Collaborate cross-functionally with product, infra, and operations to deliver robust production systems with clear SLOs and runbooks.
• Uphold high code quality through reviews, benchmarks, reproducible performance tests, and documentation.

How You Will Work

• You value measurement over assumption, using benchmarks and profiles to drive decisions.
• You communicate tradeoffs clearly (throughput vs. latency, CPU vs. memory, complexity vs. resilience).
• You write clear documentation, reproducible test harnesses, and actionable runbooks.
• You collaborate closely, give and receive constructive code reviews, and mentor peers.

Minimum Qualifications

• Expert-level proficiency in at least one of: C, C++, or Java.
• Excellent understanding of Linux kernel internals, including:
  • Memory management and allocators.
  • Device driver model and driver interactions.
  • Scheduler behavior and tuning.
  • Filesystems and VFS concepts.
• Depth in at least one of the following tracks (pick one or more):
  1. High-performance networking:
     • Strong practical experience with UDP (multicast and unicast) and TCP networking.
     • Socket options, buffer sizing, epoll, busy polling, NAPI, NIC queues, and RSS.
     • Familiarity with kernel-bypass mechanics (for example, DPDK, netmap) and XDP.
  2. Low-latency engineering:
     • End-to-end latency optimization, jitter reduction, and deterministic execution.
     • CPU pinning, interrupt affinity, NUMA, cache friendliness, lock-free or wait-free techniques, careful memory management, and lightweight logging.
  3. Distributed systems (sequencer and virtual synchrony patterns):
     • Sequencer-based total order broadcast, membership, failure handling, and consistency guarantees.
     • Tradeoffs between latency, throughput, ordering, and availability in practical systems.
• Strong debugging and profiling skills on Linux, including tool-driven investigations.
• Solid understanding of concurrency, synchronization primitives, and memory models.

Preferred Qualifications

• Experience in financial markets and trading infrastructures (for example, market data, order routing, exchange connectivity, FIX or exchange-native protocols).
• Familiarity with time synchronization for low-latency environments (for example, PTP).
• Experience with eBPF and tracing for observability in production.
• Knowledge of RDMA or kernel networking internals beyond the socket API.
• Experience writing or maintaining kernel modules or device drivers.
• Familiarity with NIC offloads and tuning (TSO, LRO, RFS/RPS, interrupt moderation).
• Experience with deterministic GC tuning (if using Java) and low-latency JVM practices.
• Benchmarking methodology knowledge: workload design, repeatability, variance analysis, and flame graph interpretation