Machine Learning Researcher - Build Models That Move Markets

Thurn Partners Ltd; New York (NY); On-site; USD 180,000 - 300,000; Full time

Job summary

Thurn Partners Ltd in New York is seeking a senior machine learning researcher to join a centre of quantitative trading excellence. You will build and test models that directly influence trading decisions, working with top researchers and traders.

The role demands leadership experience, deep Python/C++ and PyTorch skills, and a track record of deploying models that move P&L. You’ll operate on petabyte-scale data with fast feedback cycles in a highly rigorous research environment.

Qualifications

• Senior-level experience with a demonstrable impact on models and trading outcomes.
• Strong proficiency in Python, C++, and PyTorch.
• 6+ years in software or research roles, preferably in quantitative finance.

Responsibilities

• Develop and improve ML models; judged by ability to move P&L.
• Collaborate with Portfolio Manager to shape alpha-rich models.
• Lead or closely work with a team, bringing ideas from concept to tested model.
• Test ideas with rapid feedback in a fast-paced, real trading environment.
• Work with petabyte-scale proprietary data within the firm.

Skills

Leadership

Python

C++

PyTorch

Experience 6y+

MS/PhD

US work eligibility

Publications/OSS

Finance domain exp

Education

PhD or MS in CS/ML/Statistics

Tools

PyTorch

Job description

A quantitative trading firm is hiring a senior Machine Learning Researcher to work at the centre of financial trading itself, analyzing systematic trading data and building models alongside some of the strongest technical talent in the industry. The team you'd join is built from people who came out of serious research environments, PhDs and postdocs from places like MIT and other top research universities, who made the leap from academic or industrial research into quant trading and now apply that same rigour to markets. This is a hands-on role for someone who has already led real work, not just contributed to it, where every idea gets tested against hard problems and judged by one standard alone: does it move P&L. If you want to see your models shape real trading outcomes rather than sit in a benchmark, and want to do that work alongside people who've made that same jump themselves, this is that kind of seat.

What you'll do:

• Develop and improve machine learning models as part of a small, technical team, with your work judged on one clear standard: whether your model moves P&L.
• Work closely with a Portfolio Manager to shape models that sharpen alpha and improve trading outcomes, if you come from financial services, though this experience is preferred, not required.
• Lead or work closely alongside a team, bringing your own ideas and driving them from concept through to a working, tested model.
• Test ideas against real feedback quickly, this is not an environment where a model waits months to find out if it worked.
• Work with proprietary data at petabyte scale that exists only within this firm, unseen by any outside model or team.

Your profile:

• Senior-level experience, either leading a team directly or working in close technical partnership with one, with a track record of models you've built that had real, demonstrable impact.
• Strong proficiency in Python, C++, and PyTorch.
• At least 6 years working in the software industry.
• A PhD or Master's degree in computer science, machine learning, statistics or a related quantitative field.
• Must be eligible to work in the United States.
• Published research, open-source contributions, or other public evidence of original technical work, though shipped, unpublished results carry equal weight.
• Experience in the financial sector, ideally having worked closely with a Portfolio Manager or trading desk to develop models that sharpened alpha or directly improved P&L.

Why this role:

Competitive compensation, a feedback loop measured against real trading outcomes rather than a slow release cycle, and access to petabyte-scale proprietary data no other firm or model has ever touched. This is technical leadership on models that matter immediately, not research that sits in a queue.