Research Scientist, STEM

This is a research position within a STEM-focused organization working on novel ways to evaluate, train, and improve frontier AI systems. The work targets open problems where the right benchmark, dataset, or methodology does not yet exist, requiring scientists to identify gaps, formulate precise research questions, and design rigorous programs to answer them. Core areas include frontier STEM evaluation, synthetic data, hallucination and reliability, and agentic science.

• Identify high-impact gaps in benchmark and evaluation literature and design novel benchmarks within and across STEM fields.
• Develop evaluation methodologies that handle task generation, grading, contamination control, difficulty calibration, and validation.
• Build methods for generating high-quality synthetic STEM training data and study how task selection, difficulty, diversity, and filtering affect downstream performance.
• Investigate hallucination, uncertainty, calibration, and epistemic failure in technical domains, and develop methods for factual reliability, self-correction, and verification.
• Research long-horizon scientific agents that combine literature search, coding, simulation, tool use, and iterative reasoning.
• Explore new research directions in reasoning, AI-for-science, model evaluation, data generation, and emerging capabilities.

• PhD or equivalent research experience in a highly technical field such as machine learning, computer science, mathematics, physics, chemistry, biology, engineering, or statistics.
• Demonstrated ability to formulate and execute original research.
• Strong understanding of modern large language models and the frontier AI research landscape.
• Excellent experimental design, analysis, and technical communication skills.

• Compensation range of $150,000 to $300,000 OTE plus equity.
• Opportunity to publish at leading conferences such as ICLR, ICML, and NeurIPS.
• High autonomy and rapid iteration with meaningful impact on frontier AI research and enterprise deployment.