Développeur de logiciels (Recherche)

DRW Montreal•Montreal, QC

About The Position

DRW is a diversified trading firm with over 3 decades of experience that brings together sophisticated technology and exceptional people to operate in markets around the world. We value autonomy and the ability to pivot quickly to seize opportunities, which is why we operate using our own capital and trading at our own risk. Based in Chicago with offices in the US, Canada, Europe, and Asia, we trade diverse asset classes including fixed income, ETFs, equities, currencies, commodities, and energy on major global markets. We have also leveraged our expertise and technology to expand into three non-traditional strategies: real estate, venture capital, and crypto assets. We operate with respect, curiosity, and an open mind. People who thrive here share our belief that it's not just what we do that matters – it's how we do it. DRW is a place of high expectations, integrity, innovation, and a willingness to challenge the consensus. We build the systems that transform real-world events into actionable signal, with low latency and high reliability, in a trading environment where being right a second too late is the same as being wrong. This role sits at the intersection of research and infrastructure: you will form hypotheses, validate them with rapid analysis, and then build the production systems that leverage them. We also integrate LLMs as a first-class component of the toolchain, not as an afterthought experiment. People who succeed here wield an agentic harness as naturally as a profiler, and know precisely how far each remains reliable.

Requirements

3-5 years of experience designing real-time or data-intensive systems (we favor trajectory and depth over exact years).
Deep mastery of LLMs, not just usage — understanding of the real behavior of these models (context and failure modes, tool use, structured outputs, cost/latency tradeoffs, hallucination mitigation) and the ability to build the framework around a model: tools, memory, retries, evaluations, "human in the loop" boundaries. Present a concrete project you deployed and can explain end-to-end, including what broke.
Strong engineering fundamentals in Python, Go, or Rust. Preference, not exclusion — deep systems experience in another serious language is transferable.
Deep knowledge of network programming and protocols — TCP/UDP/IP, DNS, BGP, HTTP(S), WebSocket, QUIC.
Ability to rapidly turn POCs (proofs of concept) into production — validate statistical significance, iterate fast, deliver, and explain results to technical and non-technical audiences.
Bachelor's or Master's degree in Computer Science, Data Science, Mathematics, or a related discipline.

Nice To Haves

Expertise in the current LLM ecosystem — agent frameworks, MLOps, RAG architectures, evaluation tools, or local/self-hosted inference.
Demonstrated success reducing latency on NGINX, Envoy, or custom load balancing layers.
Experience in high-pressure domains — trading floors, esports, competitive sports, or hackathons.
Open-source contributions, technical blogging, or conference presentations.

Responsibilities

Extend our event-driven strategy to equity flows, market news, social media, and infrastructure telemetry.
Launch rapid analyses in Python/Jupyter to evaluate signal quality, and transform your findings into system improvements.
Design and operate agent frameworks — tools, context management, evaluations, guardrails — that actually work on our data and infrastructure, and support their quality in production.
Apply LLMs where they provide a real advantage (extraction, classification, triage, research acceleration) and decide not to use them when they are not a good fit.
Be responsible for full-stack performance, from the edge network to in-memory stores — instrument, monitor, and debug production systems alongside operations, maintaining tight SLOs (service level objectives).
Lead greenfield initiatives, design reviews, and post-mortems.