We're hiring
Work on what
happens next.
About us
Yarrow builds AI that forecasts the future — and keeps a public score. It puts probabilities on real-world events: policy decisions, market moves, geopolitical outcomes. Every forecast is locked before the answer is knowable, published, and graded when reality resolves. The same engine runs policy simulations in which AI actors respond to change inside real economic models, so impacts are computed rather than narrated.
Our mission is to help institutions make better decisions under uncertainty. Incubated by Manta, we are a small team of researchers and engineers already working with an institutional partner — and we hire on what you have built, not on titles.
Open positions[02]
01 / Research Research ScientistAI Forecasting
Research with a scoreboard — every idea you have gets graded by reality.
What you'll do
- Build forecasting systems on top of LLMs
Design how models gather evidence, reason and commit to calibrated probabilities on open, real-world questions.
- Run experiments that reality can grade
Time-frozen evidence, locked forecasts, ablations and controls — and the judgment to tell a real improvement from noise.
- Simulate how the world responds
Build multi-agent simulations of how institutions, markets and populations react to events and policies.
- Make the models better forecasters
From evaluation and data to training methods — push on what actually makes a model forecast better.
- Publish what you find
Write up results — including negative ones — and contribute to papers, benchmarks and our public track record.
What you'll bring
- Strong technical foundation (core)
A PhD or Master's, completed or in progress, in computer science, AI, mathematics, statistics, physics or financial engineering — or equivalent, demonstrated ability.
- Evidence of exceptional ability — any one of
Medals in olympiads or programming contests (IMO, IPhO, ICPC and similar); first-author papers at NeurIPS, ICML, ICLR, ACL or similar; or hands-on experience training or post-training large models.
- Strong engineering
Fluent in Python and PyTorch; you build and run experiments end to end, often with AI coding agents.
- Probabilistic thinking
At home with probability, calibration and statistics; you know why a single accuracy number can hide what matters.
- Scientific honesty
You would rather report a result that holds up than one that looks good, and you are comfortable being graded.
- Clear writing in English
You can explain a method and a result clearly to researchers and to clients.
Nice to have
- Forecasting, quant or FinTech
Quantitative research, time-series modelling, prediction markets or financial forecasting.
- Large-model training at scale
Took part in pre-training or post-training (SFT, RL) of large language models.
- Agents and evaluation
Built LLM agents, evaluation suites or benchmarks — forecasting benchmarks especially.
- Feel for the domain
Real interest in economics, policy, geopolitics or markets — the questions we forecast.
What we offer
Research that is graded by reality, not by reviewers. A small team where your ideas go straight into a product already used by an institutional partner, freedom to propose and test your own methods, and room to publish what you find.
This job description is for recruitment purposes only; final details subject to direct discussion.
02 / Business
Business Development
& MarketingFinancial Markets
Take AI forecasts that score themselves to traders, researchers and investors.
We are now taking Yarrow into financial markets, and we are looking for someone who knows how finance professionals think to bring it to them.
What you'll do
- Open doors in financial markets
Identify and engage trading desks, research teams, asset managers, brokers and financial data platforms; take opportunities from first conversation to pilot.
- Turn forecasts into financial use cases
Work with our researchers to package Yarrow's forecasts and policy simulations into use cases traders and analysts will actually use — macro and policy events, market-moving news, risk scenarios.
- Build Yarrow's voice in finance
Write research-style content — market notes, case studies, forecast reviews, including the ones we got wrong — and represent Yarrow in client meetings, roundtables and industry events.
- Bring the market back to the product
Turn client feedback on formats, timing and depth of explanation into clear input for the product roadmap.
- Shape the financial playbook
Help define packaging, pricing and go-to-market for the financial vertical as we build it.
What you'll bring
- Traditional finance background (core)
Professional trading experience (prop, buy-side or market-making desk) or sell-side research at a broker or bank. You know how finance professionals consume and judge information.
- You want to do BD or marketing
This is a commercial role. You want to open doors, present and follow through — not only do research.
- Clear research and writing
You can turn complex analysis into clear, credible content for a professional audience.
- Client-facing communication
Comfortable presenting to portfolio managers, traders and heads of research, and building relationships over time.
- Professional English
Able to write and present in English with professional clients.
- Credibility over audience size
We care about professional credibility, not follower counts. A social following or a crypto background is not required.
Nice to have
- Understanding of AI systems
A working understanding of how LLM-based systems are built and evaluated — AI harnesses, evals, calibration. No coding required.
- Feel for forecasting
Event-driven or macro trading, prediction markets or quantitative forecasting; comfort thinking in probabilities.
- A track record of public writing or speaking
Published research notes, newsletters, podcasts or conference talks — quality of thinking matters, not reach.
- Strong academic background and network
A strong degree in finance, economics or a quantitative field, and relationships across banks, brokers and funds in Asia.
What we offer
Help build Yarrow's financial vertical from the ground up — the use cases, the content and the first client relationships — alongside researchers and engineers, on a product that is scored against reality.
This job description is for recruitment purposes only; final details subject to direct discussion.