Extropic is seeking junior ML scientists to join our residency program, available on a part-time or full-time basis. Our hardware significantly accelerates certain types of probabilistic inference, and residents will help advance the science of training models within the thermodynamic framework.

### Responsibilities
– Work with senior researchers to develop the theory behind new probabilistic models and their learning methods, including energy-based models and diffusion models
– Expand and optimize our experimentation infrastructure across the model design space
– Build, visualize, and assess new architectures, training algorithms, and benchmarks
– Publish research, contribute to open source, and share design insights with our hardware team

### Required Qualifications
– Experience with scientific Python
– Experience with JAX or a comparable deep learning framework (PyTorch, TensorFlow, or Keras)
– Strong understanding of probability and linear algebra
– Projects or publications demonstrating hands-on experience in applied machine learning and data science
– Familiarity with deep learning theory and literature, including over-parameterization theory and scaling laws

### Preferred Qualifications
– Experience training energy-based models (EBMs) or diffusion models
– Experience with graph neural networks (GNNs) or graph message passing algorithms
– Experience building infrastructure for deep learning experimentation and training (Slurm, Ray, Kubernetes, Weights & Biases, etc.)
– Strong theoretical background in information geometry
– Solid knowledge of computational Bayesian methods, including MCMC sampling and variational inference
– Publications in leading ML conferences (NeurIPS, ICML, ICLR, CVPR, etc.)

Salary and equity compensation will vary based on experience. Extropic is an equal opportunity employer.