Member of Technical Staff at Mantic

You will push the frontier of forecasting accuracy by developing and refining LLM agents. Working in short, high-impact cycles, you'll improve agent scaffolding, conduct research experiments, and build internal benchmarks. This role focuses on enabling LLMs to express complex probability distributions and reason through ambiguous, real-world data to drive radical improvements in global decision-making.

• Design and implement advanced scaffolding for LLM agents to improve reasoning and research capabilities.
• Create and maintain internal benchmarks to evaluate forecasting accuracy across diverse domains like geopolitics and economics.
• Analyze model failures and bad predictions to iteratively refine how AI systems express complex probability distributions.

• Possesses exceptional coding ability and a strong background in conducting technical research projects, particularly with LLM agent frameworks.
• Demonstrates deep comfort with statistics and the ability to apply quantitative reasoning to uncertain, judgmental forecasting problems.
• Brings a high-performance work ethic and evidence of exceptional achievement, likely proven through experience in AI labs, quant finance, or academic research.

===

• Work at the frontier of AI forecasting with a system already recognized by Polymarket as the best-performing AI in the field.
• Collaborate with an elite technical team featuring alumni from Google DeepMind, Citadel, Goldman Sachs, and top-tier academic institutions like Oxford and Cambridge.
• Join a well-funded mission with a recently closed $23M seed round following a $4M pre-seed, providing significant runway and growth potential.