Quant Research Scientist, Inference on Novel Data ($500k-1.5m TC)

Thurn Partners; New York (NY); On-site; USD 120,000 - 180,000; Full time

Job summary

Thurn Partners is building a central research team inside an advanced systematic fund. The group tackles questions with limited data, such as determining if an industrial process is running above or below normal worldwide this month, using satellite imagery, vehicle movements, and multilingual filings.

They value original research (a doctorate is the usual form) and recognize other achievements like first‑author papers, influential libraries, or notable competition results.

Job description

Inference on Data Nobody Has Modelled Before

Here's an example problem this group solves:

Someone wants to know whether a particular industrial process is running above or below its normal rate, worldwide, this month. Nobody has ever measured it and is no series to extend and no label to train it against. What exists is satellite imagery on an inconsistent revisit schedule, a vendor feed of vehicle movements with unexplained gaps, and a decade of filings in four languages. Decide whether the question can be answered at all, then answer it, then establish that the answer is real.

Nobody can mark your work, because if the ground truth existed somewhere, someone would already be trading on it.

The group is a central research team inside an advanced systematic fund. Working with every investment team in the firm rather than sitting under one desk. Mostly not from finance: physicists, astrophysicists, materials scientists, statisticians who moved into applied work. Iteration runs weeks to months per question, and a negative result you have properly established counts as an output. A separate engineering team takes models into production, so the pipeline does not become yours by default.

What they look for:

• Evidence that you can do original research. A doctorate is the usual form of that evidence and most of the team has one.
• It isn't the only form: a first-author paper written before you had a PhD, a library other people depend on that you wrote yourself, an olympiad or ICPC record, a strong placing against a serious field are read the same way.

From there:

• Code that runs on someone else's machine. Python ideally.
• The judgement to decide for yourself which question is worth three months and which is worth an afternoon
• No finance background required or expected

If you are not looking, say so and and that's fine. This advert is partly for those who are rarely on the market.