FutureSearch is looking for exceptional Software Engineers to evaluate and improve state-of-the-art forecasting and agentic LLM web research.

We are an elite team of engineers and researchers at the frontier of AI epistemics. We have the best publicly available forecaster, with a public track record in forecasting tournaments at evals.futuresearch.ai and markets.futuresearch.ai. We publish papers, such as our ICLR Workshop paper in Jan 2026, and benchmarks like Deep Research Bench and Bench to the Future. We use our frontier research tools to research the future, such as co-authoring the AI 2027 Timelines Forecast. We work with frontier labs on research, evaluation, and training.

This is a senior role with wide scope. In a typical month you might ship a product feature in the FutureSearch app, extend the pipelines behind our benchmarks, and work out why an agent pool times out under production load. You own what you build end to end: design, deployment, monitoring, and the debugging when it misbehaves. We're a small team (11 people), so we hand things off whole and expect them to get done.

Requirements

You're someone who can take a rough idea, like "researchers should be able to replay any production forecast against a modified pipeline and see what changed", and make it real: a working prototype this week, then the hardening that turns it into something the team relies on. Much of that building happens through coding agents, so the leverage is in what you choose to build and how you verify it. You either have experience with our stack, or can pick it up fast: Python on GCP and Azure, Kubernetes, Ray, Postgres, ClickHouse, Dagster. You are fascinated by AI and want a job where you push what's possible. You work quickly, and love using the products that you build.

The strongest candidates will have experience with:

1. Owning production backend systems end to end, including deploying, monitoring, and debugging them when they misbehave
2. Building on LLM APIs and working daily with coding agents like Claude Code, with good instincts for when their output is wrong
3. Research, in academia or in industry. Many of us come from research backgrounds, and engineers who read papers and prototype the ideas in them thrive here

The ideal candidate has a burning desire to improve AI epistemics in a way that improves humanity’s decision making in the near-term. Some of the best fits for this role are people who nearly became research scientists and chose engineering instead.

Benefits

FutureSearch is fully remote. SF, NYC, and London are our primary in-person locations, and we travel to meet 4-6 times a year. We have flexible hours, and operate in a high trust environment.

Our salary range is $125,000 to $225,000, varying primarily by country. We offer higher-than-average equity. You should expect a total compensation package that is near the top for any seed-stage startup, with large room to grow.

We offer full benefits, varying based on your country, including: healthcare, paternity/maternity leave, equipment, and travel expenses.