Always Innovating

C++ Engineer

Volant Trading was founded in 2006, and is one of the elite High-Frequency Trading firms doing FPGA-hardware-accelerated ultra-low-latency trading, where a handful of nanoseconds (literally) can make the difference between being profitable and not. Our trading systems are written in C++ and run on Linux, interfacing with and controlling the FPGA designs written in System Verilog.

Being consistently successful is not the result of one or two systems working well – a plethora of things must be operating well concurrently, each with their own blend of optimisation criteria – for latency, throughput, developer productivity, maintainability, agility, testability, robustness, operational supportability, and all the properties that support those like concision, self-documenting expressiveness, modularity, and abstraction/encapsulation. This requires discerning, targeted application of modern C++ - and more generally software-engineering - best practices. We want people who understand – or have an aptitude for – Design Principles and Patterns, and modern C++ idioms and usage.

You might be expecting us to make the HFT-industry-standard disingenuous unsubstantiated waffling claims about innovation and a culture of excellence. Instead, here are some concrete, differentiating qualities that should actually help you decide if this is the job for you...

About the role

• Architecting/design, development, testing, deployment, operation, support and trouble-shooting of our trading systems and all in-house libraries and tooling supporting them; everything's decided and done locally in Hong Kong by the small, tight-knit, in-office team of Linux/C++ and Verilog/FPGA engineers that you'll be joining; we are not current open to remote work
• You'll not just have a spot at the table while the IT team and traders make key decisions - we hope you'll be helping set the agenda and in the thick of things, suggesting solutions for emerging business issues and opportunities
• As engineers - tempered only by common sense and practicalities - we can empower ourselves with the latest tooling: compilers (we're using GCC 13.1), coding techniques, IDEs (most of us choose CLion), devops tooling (we've been an Atlassian shop, but will trial JetBrain's Space), hardware (from getting the latest Ryzens and core-i9s as we need them, to dual-AMD-EPYC 9684Xs), Linux Distros (RHEL 9.3), 3 rd party libraries (e.g. boost 1.81), and utilities; if you know a smarter way to work, we'll want to explore it and empower you and each other
• Our existing team members and traders will help you develop an holistic understanding of how things fit together, as well as low level details, to make informed architecture, design, and implementation choices; you won't get the chance to develop this kind of understanding of a complete, mature, successful trading operation at many companies
• You don't need to be self-dependent - collaboration and knowledge sharing help us perform optimally as a team - but you will need to be largely self-motivated and self-organising, and have confidence in the impact you're making; while someone (including you) might throw a task description into a JIRA or similar if organisationally helpful to the interested parties, or required for Compliance with regulatory, exchange- or broker-specific policy, there are no IT or project managers wanting to be between you and your users, or breaking down your tasks into week-sized chunks and JIRA subtasks that get ticked off like homework with a pat on the back
• Similarly, there are no quants writing financial libraries; no testing teams, project managers, operations or N-th level support, database admins, approval committees or advisory boards, R&D or procurement teams; no other team responsible for fixing or enhancing library "X" that you'll have to wait for; we do everything, everywhere; you'll do everything; you'll need to want to, like us
• This job requires actual shut-out-the-world old-fashioned thinking - sometimes systematically, sometimes creatively, and a practical ability to optimally model problem domains in code; stackoverflow and chatgpt won't carry anyone here
• We have an unapologetically tough candidate process – we need to hire better people than our competitors to evolve faster than them, so we must screen and select more discerningly; feel free to be tough on us too - we want people who want good colleagues