Member of Technical Staff at Mantic

You will join a world-class technical team to solve judgmental forecasting using LLM agents. By building deep research agents and experimenting with probability distribution tools, you will improve decision-making in geopolitics and economics. This role involves rapid experimentation cycles and pushing the frontier of AI accuracy alongside veterans from DeepMind and Citadel.

• Design and implement LLM agent scaffolding to improve forecasting accuracy across geopolitics, business, and technology sectors.
• Develop new benchmarks and tools that allow LLMs to express complex probability distributions and conduct deep autonomous research.
• Conduct short, high-impact experiments to analyze prediction failures and iterate on the system's reasoning capabilities within one-week project cycles.

• Possesses a strong work ethic and exceptional coding ability with a proven track record of delivering high-quality research or technical projects.
• Demonstrates comfort with statistics and probability, ideally with experience in quant finance, AI labs, or advanced academic research on LLMs.
• Shows evidence of exceptional achievement in any domain and is eager to work in-person 5 days per week in a high-intensity London office.

• Work with an elite technical team from Google DeepMind, Citadel, Palantir, and Oxford VGG, tackling one of the most intellectually challenging problems in AI.
• Join a project recognized by Polymarket as the best-performing AI forecasting system, recently placing 4th out of 500+ participants in the prestigious Metaculus Cup.
• Impact a mission-driven product featured in The Atlantic and The Guardian that aims to radically improve global decision-making through superior predictive reasoning.