Empire Postdoc Fellow in Self Improving AI

reinforcement learning and post-training, and reliable mathematical problem solving. Potential research directions include: • Foundations of self-improvement and reasoning. Understand how pretraining, supervised fine-tuning, and reinforcement learning shape reasoning capabilities; when self-supervised training generalizes to harder problems; and what limits sustained improvement. • Learning from weak and verifiable feedback. Study weak-to-strong generalization, model-generated supervision, and verification-guided learning, with attention to reasoning faithfulness, reward exploitation, and preservation of existing capabilities. • Data for sustained improvement. Design synthetic-data generation, selection, curricula, and replay methods that turn computation into useful learning signals while limiting bias, overfitting, forgetting, and model collapse. • Applications in AI for mathematical discovery. Develop methods for proof search, conjecture generation, abstraction, counterexample search, and constructive mathematical problems. Combine learned models with symbolic tools and automated verification, where available, to learn reusable problem-solving strategies. The Fellow will use Empire AI for large-scale, reproducible experiments that connect these questions to practical algorithms. The work will include efficient adaptation, compute allocation across generation, verification, and training, and criteria for stopping or rolling back harmful updates. Expected contributions include research publications and reusable, open-source pipelines, evaluated for both reasoning performance and improvement per GPU-hour. The research agenda will be developed jointly with the candidate and may emphasize theoretical foundations, algorithm design, or empirical discovery. The candidate will collaborate with both hosts and participate in the NYU research community. Annual base salary: starting salary range: $62,500-92,500. The Empire AI fellowship provides salary and fringe support for two years. Candidates must hold a Ph.D. in mathematics, computer science, machine learning, statistics, data science, or a related discipline by the appointment start date. We seek researchers with a strong record in machine learning, optimization, or a closely related area; the ability to conduct rigorous computational experiments; and strong communication and collaboration skills. Relevant experience in one or more of the following areas is particularly welcome: ? Reinforcement learning, language-model post-training, planning, or search. ? Learning theory, optimization, generalization, or learning from limited or imperfect supervision. ? Synthetic-data generation or selection, efficient adaptation, or reproducible GPU experimentation. ? AI for mathematics, automated theorem proving, verification, reasoning evaluation, or AI for scientific discovery. Applicants with either a theoretical or an empirical research emphasis are encouraged to apply. Experience across all of these areas is not required. Please submit a single PDF through the Interfolio application link containing: ? A curriculum vitae, including a publication list. ? A research statement outlining previous work, future directions, and connections to this position. Please indicate your preferred start date in 2027-2028 academic year. If needed, the start date can be modified to earlier 2027. ? Up to three representative publications or preprints. ? Contact details for three referees. Review of applications will begin on December 1, 2026 and continue until the position is filled. For people in the EU, click here for information on your privacy rights under GDPR: www.nyu.edu/it/gdpr NYU is an Equal Opportunity Employer and is committed to a policy of equal treatment and opportunity in every aspect of its recruitment and hiring process without regard to age, alienage, caregiver status, childbirth, citizenship status, color, creed, disability, domestic violence victim status, ethnicity, familial status, gender and/or gender identity or expression, marital status, military status, national origin, parental status, partnership status, predisposing genetic characteristics, pregnancy, race, religion, reproductive health decision making, sex, sexual orientation, unemployment status, veteran status, or any other legally protected basis. All interested persons are encouraged to apply for vacant positions at all levels. Sustainability Statement NYU aims to be among the greenest urban campuses in the country and carbon neutral by 2040. Learn more at nyu.edu/sustainability