Member of Quantitative Staff

Qualified Intelligence is building cost of corrections insurance for AI work product: when work done with AI turns out wrong, we cover what it costs to fix. It is first-party coverage, triggered by a documented correction rather than a third-party claim.

There is no existing class of business for this risk, so most of the quantitative questions are still open. How likely is it that a given piece of AI work needs correcting? When it does, what does the cost distribution look like, especially out in the tail? And how correlated are those losses when thousands of insureds depend on the same handful of models and one of them quietly gets worse? We are hiring a Member of Quantitative Staff to work on questions like these.

The work is applied research. You will define the exposure, pull and study whatever data we have, build models, and test them against real underwriting decisions. What you find feeds straight into how we select, price, limit, and decline accounts, so you will work closely with underwriting and your analysis will be used quickly. As the book grows, you will turn one-off analyses into tools underwriters can use on their own.

This is an early career role, and you do not need insurance experience. What you do need is to be steady when the data is thin and the right model is not obvious, when the honest move is to reason from first principles and be explicit about what you are assuming.

What you'll do.

• Build models for the pieces of this risk: exposure, frequency, severity, and how losses aggregate.
• Work with underwriting to decide what a submission should ask for, and evaluate submissions and claims as they come in.
• Test pricing logic against the evidence, and write up what holds and what doesn't.
• Build the tooling that turns your analysis into something underwriters can apply consistently.

What we look for.

• Real fluency in probability and statistics. You can reason about distributions and dependence, and you know what your model is quietly assuming.
• Enough programming to get answers out of messy data yourself. Python is the most useful language here, but what matters is that you can build a simulation, check a model, and explain the result.
• Curiosity that doesn't switch off. If a number looks wrong, you need to know why.
• Writing clear enough that an underwriter can act on it.

Backgrounds that tend to fit: quantitative finance, trading research, risk modeling, or insurance analytics; a PhD, postdoc, or strong academic record in statistics, economics, computer science, physics, mathematics, engineering, operations research, or data science; or data engineering, applied AI, and production machine learning on systems that use models in real workflows.

Show us your work. We care most about evidence that you can take an open-ended problem and drive it to a result on your own: a thesis, a forecasting competition, a simulation, a paper you replicated, an agent you built, or anything with a clear outcome. If defining the problem was half the battle, that is the part we most want to hear about.

Compensation and benefits.

• Base salary: $225,000
• Health insurance: Fully funded medical, dental, and vision
• Paid time off: 15 days

Apply. Email hiring@qualifiedintelligence.com with a resume or LinkedIn profile and a short note about a project or paper you are proud of: what the problem was, what you did, and what you found. If you've built something with AI, send that too, and if there's a paper you keep coming back to, tell us which one and why.