About the job
Would you like to join a team curious about understanding how foundation models work and to expand their capabilities in scientific domains? We perform and publish novel research and apply our findings to drive product directions. If you like designing clever approaches to understanding complex phenomena and using that knowledge to solve practical problems then this is the position for you.
Responsibilities
Design robust experiments and methods to understand what foundation models do under the hood.
Implement your methods and designs in experiment pipelines.
Analyze and interpret your experimental results.
Communicate results to teams across Apple.
Write papers for publication in top-tier conferences/journals.
Qualifications
Minimum
PhD in computer science, statistics, physics, chemistry, electrical engineering, or operations research. Other hard sciences may also be considered.
3 publications in top-tier machine learning, statistics, or natural language processing venues.
Deep knowledge of foundation models and experience training them and applying them to real, complex datasets as demonstrated through publications or code repositories.
Experience designing experiments to understand how foundation models work.
Knowledge of Bayesian statistical methods and how they are used for scientific inference.
Preferred
Experience with large language models and – both training them and using tools like vllm for inference.
Proficient implementing ML models and experiments in Python and Pytorch/Jax.
Familiarity with interpretability methods like activation patching/causal tracing.
Demonstrate the ability to refine ambiguous research ideas to construct a coherent and logically sound story.