Finance is interesting from a machine learning perspective. It's one of the few industries in the world where you basically have an untapped, unlimited source of data. We try to predict stock prices based on data. Basically, 100% of my day involves ML and systems engineering. Basically, no finance is required. When you apply ML to finance, you're basically doing weather forecasting. Solving very interesting challenges every day on the boundary between both machine learning as well as software engineering. It's actually not that different from a lot of the frontier AI labs. We work with researchers who develop these machine learning models, so they need lots and lots of data, and lots and lots of compute, and we apply it to predicting the future of financial markets. In our team, we have people from various kinds of backgrounds. We have software engineers, for example, coming from gaming industry or, for example, we have people coming from banking. Financial knowledge is not required. Financial curiosity is definitely an asset. But what really matters are software engineering skills, statistics, linear algebra, and the foundation of machine learning. Machine learning engineers and quant developers both have very important roles, and they do overlap. The machine learning engineers, those are the guys that will sit with the quants and ensure that their workflows are fitting the best possible models and running their training the most efficiently. Whereas the quant developer is more incorporating the researcher's new models into the existing trading platform. We had an idea that we knew was valuable, but it was taking a really long time to train the models, and so I had to go away and learn how to write custom GPU code and build TensorFlow from source. Go from not knowing how to write any GPU code to writing a kernel or a GPU program that was more efficient than what was in the very public machine learning library. Every single model that comes out, it's unique in some weird and wonderful way. You feel like you've had direct involvement in delivering the new research models. That one's particularly exciting. It's actually insane how much care we have for each other. When you do something right, it can be something tiny and you just fixed it for your sake, and then I'll tell you at a team meeting, being like, "Oh, by the way, Julia fixed it." Having this access to this incredible compute resource means that we can spend longer thinking about things and less long waiting for results to come back, which is amazing. So I've got this new feature that I want to test out. I will turn it on, and then by tomorrow morning, I've got my same results and I can look at the pretty graphs and determine whether I've made money or lost money in a straight line. We look for a spark, someone who is going to be proactive in finding the right work. We have no strong preferences when it comes to any particular tools or technologies that you need to work with. It's mostly about the algorithmic thinking. We will frequently give candidates the option of what language they want to choose. I had one the other day where someone succeeded in answering one of our quant dev interview questions in C++, so you can do it. I found the interview process really comprehensive. Multiple stages, and I look at it as a positive because everybody who had to go through a similar process is going to be quite skilled and very nice to be working with. It's very hard for me to imagine a place with a higher concentration of clever, incredibly talented people. I learn from each of them every day. I admire all of them. You shouldn't really be afraid to apply to G Research. You will be able to work with people who are very passionate about trying to solve this problem, of trying to create the best predictor in the entire world with such motivated and brilliant people.