Systems Engineer - HFT | C++ | New York | $500k-1m

A global market making firm is building a new US equities and ETF business in New York and hiring the founding engineering team. The architecture does not exist yet. Whoever joins in the next few months decides the threading model, the memory layout, the wire protocol handling and how the whole thing is tested, and lives with those decisions for the next decade.

You do not need to know anything about markets. The hiring manager would rather teach an exceptional systems engineer how exchanges work than teach a trading systems person how to write software properly. Several of the strongest engineers on the desk arrived from chip companies, database internals and kernel work.

What people take this for is the feedback loop. You change something and the same day you can see what it did to the money. Nobody writes a design document for six weeks, and the team is small enough that your name is on the parts you built. The firm trades on more than a hundred exchanges globally and has been profitable across market conditions. This desk is its push to lead US equities the way it already leads elsewhere, and it is funded accordingly.

Responsibilities

• Build market data, order management and execution from scratch in modern C++ on Linux, with the core design decisions still open.
• Work on the tail rather than the mean. The end-to-end path is measured in microseconds and the hard part is what happens at the ninety-ninth percentile.
• Spend real time in profilers, thinking about cache lines, allocation, contention and what the NIC is doing.
• Own the software side of the boundary with the FPGA team rather than handing work over it.
• Sit with the traders and researchers you build for, and turn their ideas into systems that hold up in production.

Qualifications

• Three to eight years writing production C++ where performance was the point rather than an afterthought.
• Experience building the internals of something serious: an operating system or kernel, a compiler or runtime, a database or storage engine, a filesystem, a network stack, an embedded or real-time system, or accelerator kernels.
• Comfort close to the machine, whether that is kernel bypass and RDMA, driver or firmware work, GPU kernels, or low-level optimisation in another demanding domain.
• No trading or finance background required.
• Someone other engineers want in the room when a decision is being made.

New York, on site. Relocation is routine and not a barrier.