The Jane Street Visiting Researcher Program was designed to further facilitate that collaboration. The program is open to all active researchers with a doctorate. We welcome applications from folks in many stages of their academic career: pre-faculty, current faculty, and junior researchers fresh out of graduate school are all eligible to apply. We wish to learn more about visitors' research while giving them the ability to evaluate their ideas in an industrial setting. Researchers come to work at Jane Street, lending their expertise to problems we care about, and working to communicate some of those insights to the broader academic community.

Researchers can join us as a visitor for as little as three months, or up to a year, and while we enjoy working together in person, remote opportunities are also available. Below is a list of areas that we think are especially well-suited to this program.

Machine Learning. Machine learning is at the core of our business. It drives many of our most important trading strategies, and is used to automate and accelerate much of our daily work. This involves both 3rd-party models, and custom models that we train on our own infrastructure. As such, we're interested in a wide variety of problems in the ML space.

Programming languages. Over 80 million lines of our software is written in OCaml, so we work on improving many aspects of the language and the compiler that supports it. We are designing and implementing ground-breaking approaches to data-race freedom and fine-grained control of representation, all without losing polymorphism, garbage collection, or backward compatibility. We call our set of OCaml extensions OxCaml, which we release for open-source use.

Systems. The demands of trading have led us to invest in a wide variety of complex systems that cross over into many traditional areas of systems research, including distributed shared logs, incremental and graph-structured computation systems, query planning and optimization, and distributed storage systems.

Hardware. Both ASICS and reconfigurable hardware like FPGAs are an effective tool for building powerful accelerators for all sorts of applications, from packet processing to machine learning. But hardware design is slow and exacting work, and so we've built tools and abstractions that simplify and accelerate the process. We are also interested in researching how we can combine verification techniques and methodologies with Hardcaml - the language we use to implement hardware at Jane Street.

Networking. Trading makes a lot of unusual demands on the network, and this has led us to work on better approaches and abstractions for configuring and analyzing our network, all in the context of network designs that differ in important ways from the mainstream approaches to designing cloud-computing networks.

Graph-structured and incremental computation. This comes up in a variety of contexts, from evaluating financial models to designing build systems for compiling our code. We've built several systems in this space, and continue to work on improving the performance, scalability, and reliability of these systems.

That said, the list is not exhaustive, and we're happy to consider collaborations in other areas.

If you're interested, please fill out this form. Don't worry too much about every little detail. It's an informal process, and we expect to tailor the program to each individual. We'll be in touch if it sounds like there could be a good fit.

Researchers in type systems may be interested in our specific position visiting our OCaml Language team.