Research Staff, LLMs
Research and develop large language models by designing experiments, advancing transformer training and optimization, and scaling successful approaches on distributed infrastructure into production.
About the opportunity
Anyone AI is seeking Astronomy & Cosmology Experts to develop challenging tasks that test AI agentsβ scientific reasoning and research capabilities. You will create original tasks and reference solutions, then evaluate agents completing workflows using astronomical data and computational tools.
Location: Remote β Latin America and Europe
Engagement: Project-based with flexible hours
Compensation: US$30βUS$80 per hour, depending on expertise and project requirements
Responsibilities
Design tasks involving observational astronomy, astrophysics, celestial mechanics, or cosmology.
Develop reference solutions with transparent calculations, assumptions, and supporting evidence.
Evaluate agents using astronomical catalogs, data archives, simulation tools, or analysis software.
Define scoring criteria for physical reasoning, numerical accuracy, and uncertainty handling.
Identify errors in units, coordinate systems, model selection, or interpretation.
Requirements
Advanced degree in astronomy, astrophysics, cosmology, or a related discipline, or equivalent professional expertise.
3+ years of relevant research or applied experience.
Strong quantitative skills and experience working with astronomical data or models.
Fluent English, with strong written communication skills.
Preferred qualifications
Experience with Python, Astropy, scientific computing, observational data reduction, or numerical simulations.
Research and develop large language models by designing experiments, advancing transformer training and optimization, and scaling successful approaches on distributed infrastructure into production.
Develop training data, agent architectures, and evaluation methods that help language models reliably execute multi-step engineering workflows across CAD, CAE, and semiconductor/EDA.
Research and co-design algorithms that adapt in real time to product feedback, improving model efficiency and system-wide performance across the machine learning stack.
Build scientific benchmarks and evaluation tasks that help AI agents reason like expert geroscientists, covering aging biomarkers, experimental analysis, multimodal data, and therapeutic discovery.
Combine life sciences research expertise with hands-on Python skills to create and solve scientific problems, analyze biological datasets, and review AI-related research solutions and code for accuracy and reproducibility.