Member of Technical Staff at Mantic

Join an elite technical team to solve judgmental forecasting by building advanced AI systems. You will work primarily with LLM agents, designing scaffolds and running rapid experiments to improve predictive accuracy. This role offers the unique chance to outperform human superforecasters and shape decision-making in geopolitics, business, and technology.

• Develop and optimize LLM agent scaffolding to improve forecasting accuracy through rapid one-week project cycles and internal benchmarking.
• Build sophisticated deep research agents and experiment with novel tools that enable LLMs to express and refine complex probability distributions.
• Conduct detailed post-mortem analyses on incorrect predictions to identify systemic reasoning flaws and implement architectural improvements to the agent frameworks.

• Possesses exceptional coding ability and a strong work ethic, with a proven track record of delivering high-quality results in research-intensive environments.
• Demonstrates a deep comfort with statistics and experience in AI labs, quant finance, or academic research focusing on LLM frameworks and agentic behavior.
• Shows evidence of extraordinary achievement, such as top-tier academic results or success in competitive environments like math olympiads or elite engineering programs.

• Work with an elite technical team of alumni from Google DeepMind, Citadel, Oxford's VGG, and Palantir on one of the most challenging problems in AI.
• Mantic is already recognized by Polymarket and The Guardian as a world-leading AI forecasting system, recently placing 4th out of 500+ participants in the Metaculus Cup.
• You will have direct impact on the frontier of AI reasoning, transitioning from statistical models to agents that perform deep research and complex probability distribution analysis.