Skip to main content

Jane Street

Machine Learning Researcher

New York, New York, United StatesmidAdded today

About this role

Jane Street seeks Machine Learning Researchers to collaborate on novel ML techniques applied to systematic trading strategies. You'll work with experienced researchers on real market datasets and cutting-edge infrastructure while developing practical models that address challenges unique to trading environments.

What you'll do

  • Conduct exploratory analysis on large, complex datasets to identify modeling opportunities
  • Design and implement machine learning models using Python and modern frameworks
  • Collaborate with full-time researchers on end-to-end projects from conception to evaluation
  • Debug training issues, tune hyperparameters, and analyze model predictions
  • Investigate novel modeling paradigms and unconventional approaches to trading challenges
  • Synthesize insights from cross-functional colleagues with diverse expertise

What they're looking for

  • Machine learning (practical implementation and theory)
  • Python programming
  • ML frameworks (e.g., TensorFlow, PyTorch)
  • Statistical and mathematical reasoning
  • Data analysis and exploratory data science
  • Model debugging and hyperparameter optimization
  • Problem-solving in ambiguous, open-ended situations
  • Rapid prototyping and iterative development

Benefits

  • Access to petabytes of proprietary market data
  • Computing cluster with hundreds of thousands of cores
  • High-end GPU cluster for advanced modeling
  • Mentorship from experienced ML researchers
  • Educational classes on systematic trading and markets
  • Exposure to novel techniques and blue-sky research

Likely interview questions

  • Walk us through a machine learning project you've completed—what problem did you solve and what techniques did you use?
  • How would you approach debugging a model that's performing unexpectedly poorly on a new dataset?
Apply on the employer's site

Opens the official application on the employer’s site. No login required.