The role involves working in a research lab funded by a hedge fund, focusing on deep learning and machine learning research. The lab aims to optimize for research output and runway, taking the 'bitter lesson' seriously by leveraging vast amounts of compute and data. Techniques of interest include test time compute, active inference, meta-learning, self-play, and system 2 techniques. Candidates should have interest and knowhow in deep learning, transformer models, reasoning, world models, learning-to-learn, mechinterp, Bayesian methods, and probabilistic programming. Flexible options scheme based on results is included in the compensation.
