If you haven’t heard of it, Depth First Learning is a wonderful resource for learning about machine learning.

Jane Street is sponsoring this year’s MakeMIT hackathon, and we wanted to create a prize for the winners that would do justice to the maker spirit of the competition. As makers ourselves – it’s not unusual to find a “software” engineer here who hacks on FPGAs or who has a CNC machine at home – it fe...

At Jane Street, over the last few years, we’ve been increasingly exploring machine learning to improve our models. Many of us are fascinated by the rapid improvement we see in a wide variety of applications due to developments in deep learning and reinforcement learning, both for its exciting potent...

In a previous blog post we detailed how we used OCaml to reproduce some classical deep-learning results that would usually be implemented in Python. Here we will do the same with some Reinforcement Learning (RL) experiments.

This blog post is about an interesting detail about machine learning that I came across as a researcher at Jane Street - that of the interaction between L2 regularization, also known as weight decay, and batch normalization.

At Jane Street, our web UIs are built on top of an in-house framework called Incr_dom, modeled in part on React’s virtual DOM. Rendering different views efficiently in response to changes made to a shared model is a quintessentially incremental computation—so it should be no surprise that Incr_dom i...

At Jane Street, we often work with data that has a very low signal-to-noise ratio, but fortunately we also have a lot of data. Where practitioners in many fields might be accustomed to having tens or hundreds of thousands of correctly labeled examples, some of our problems are more like having a bil...

Last year we held a machine learning seminar in our London office, which was an opportunity to reproduce some classical deep learning results with a nice twist: we used OCaml as a programming language rather than Python. This allowed us to train models defined in a functional way in OCaml on a GPU u...

Yet again, intern season is coming to a close, and so it’s time to look back at what the interns have achieved in their short time with us. I’m always impressed by what our interns manage to squeeze into the summer, and this year is no different.

With the external release of OCaml 4.07.0 imminent, we in Jane Street’s Tools & Compilers group have been planning what we want to work on for inclusion in OCaml 4.08. These days OCaml uses (or at least attempts) a time-based release process with releases scheduled every 6 months. We’re trying to av...

Expect tests are a technique I’ve written about before, but until recently, it’s been a little on the theoretical side. That’s because it’s been hard to take these ideas out for a spin due to lack of tooling outside of Jane Street’s walls.

One of the joys of working at Jane Street for the last 15 or so years has been seeing how our software stack has grown in scope. When I started, I was building pretty narrowly focused systems for doing statistical research on trading strategies, and then building systems for executing those same str...

Imagine a system for editing and reviewing code where:

Interested in learning OCaml? In the NYC area? Then this might be for you!

People often think of formal methods and theorem provers as forbidding tools, cool in theory but with a steep learning curve that makes them hard to use in real life. In this post, we’re going to describe a case we ran into recently where we were able to leverage theorem proving technology, Z3 in pa...

As Jane Street grows, the quality of the development tools we use matters more and more. We increasingly work on the OCaml compiler itself: adding useful language features, fine-tuning the type system and improving the performance of the generated code. Alongside this, we also work on the surroundin...

This post is aimed at readers who are already familiar with stochastic gradient descent (SGD) and terms like “batch size”. For an introduction to these ideas, I recommend Goodfellow et al.’s Deep Learning, in particular the introduction and, for more about SGD, Chapter 8. The relevance of SGD is tha...

It’s time for our next Jane Street Tech Talk. When we’ve solicited suggestions for topics, one common request has been to talk about our internal development process. Our next talk, How Jane Street Does Code Review, should fit the bill. The talk is being given by our own Ian Henry, and discusses how...

After a summer hiatus, the Jane Street Tech Talks series is back on for the fall! Last we left it, our very own Dominick LoBraico presented on the evolution of our internal configuration methodology and the systems that support it. For anybody that missed it, you can check out a recording of the tal...

Trading is a competitive business. You need great people and great technology, of course, but also trading strategies that make money. Where do those strategies come from? In this post we’ll discuss how the interplay of data, math and technology informs how we develop and run strategies.

For those of you interested in what what interns do at Jane Street, here’s a post from former intern Tristan Hume, on his work developing tree-diffing algorithms last summer at Jane Street. It’s a fun (and very detailed!) read.

People seem to enjoy talking about programming methodologies. They give them cute names, like eXtreme programming, Agile, and Scrum; run conferences and build communities around them; write books that describe how to use them in excruciating detail; and manifestos that lay out their philosophy.

Jane Street is looking to hire an engineer with experience in both software and hardware design to work on FPGA-based applications, and on tools for creating such applications.

Intern season is coming to a close, and it’s a nice time to look back (as I’ve done in previous years) and review some of what the interns did while they were here. The dev intern program has grown considerably, with almost 40 dev interns between our NY, London, and Hong Kong offices.

There are abundant resources online trying to scare programmers away from using shell scripts. Most of them, if anything, succeed in convincing the reader to blindly put something that resembles

Update: I’m excited to say that we’ve now hired a (great!) technical writer, so the position is closed.

We have a new tech talk coming up on May 17th, from our very own Dominick LoBraico. This one is about how to represent configurations with programs. In some sense, this is an obvious idea. Lots of programmers have experienced the dysphoria that comes from watching your elegant little configuration f...

It’s often surprising just how much software performance depends on how the software is deployed. All the time and effort you’ve invested in optimization can be erased by a few bad decisions in scheduler policy, affinity, or background workload on a server.

From now and then, I found myself having to write some mechanical and repetitive code. The usual solution for this is to write a code generator; for instance in the form of a ppx rewriter in the case of OCaml code. This however comes with a cost: code generators are harder to review than plain code ...

I’m happy to announce our next public tech talk, called Seven Implementations of Incremental, on Wednesday, April 5th, presented by yours truly. You can register here.