Jane Street builds trading models that adapt to regime shifts in noisy data
Their ML group runs distributed training on thousands of GPUs while dealing with low-latency requirements and the feedback loop of their own trades influencing the data. Researchers and engineers sit together to tune hyperparameters, debug CUDA kernels, and productionize models that must handle sudden structural changes from elections or pandemics. The scale (1+ exabytes storage, $400B daily volume) highlights real constraints on inference speed and model robustness that matter when you're wiring agents into production pipelines.
janestreet.com
Dwarkest interview with Machine Learning :: Jane Street
Jane Street is a quantitative trading firm and liquidity provider with a unique focus on technology and collaborative problem solving.