Skip to content
Careers / Open role

Quantitative Underwriting Engineer

Location
New York City · In office
Total cash compensation
$225,000–$300,000 + equity
Reports to
President
Role
Individual contributor

Understand the home.
Price the risk.

Work at the frontier of physical risk modeling, where machine learning, statistical inference and physics meet real underwriting decisions.

This role is for you if

  • You want to solve problems without an established modeling playbook.
  • Your models have informed decisions with real money at stake.
  • You build models in production code and test them against real outcomes.

Where you might be coming from

Catastrophe modeling, reinsurance, insurance-linked securities, mortgage credit, weather or energy risk.

Apply for this role Explore the role
Hover to see what holds it together. Tap or press Space to toggle.Illustrative framing from roof to foundation, not a verified construction drawing.
An illustrative day on the job

The storm passed.
Was the model right?

Before landfall, you estimated the damage to each home. Now the claims are arriving. Every difference is something to investigate.

  1. 01 Reconstruct the wind and hail each home experienced.
  2. 02 Compare damage and loss with the prediction.
  3. 03 Trace discrepancies to the data, assumptions or model.
  4. 04 Test the correction against the approach it replaces.
  5. 05 Update pricing, selection and mitigation where the evidence supports it.
01 / Atlas

Follow one home.

A physics-based digital twin of every home we insure. You connect what we know about the house to the risk we take.

01

Observe

Assemble imagery, inspections, permits, registries and claims.

02

Verify

Resolve conflicting sources and establish which facts we can trust.

03

Model

Simulate how the house responds to wind and hail. Estimate damage and loss.

04

Price

Set the premium and decide which homes we write, decline or refer.

05

Strengthen

Identify improvements, quantify their value and reprice verified work.

06

Reconcile

Compare predictions with outcomes and use the evidence to improve the model.

You own the full loop

The data, vulnerability models, pricing, selection and record of each home.

02 / How you improve it

A claim becomes
a better prediction.

Illustrative example

The roof lost more
than you expected.

Was the roof condition recorded incorrectly? Did the home experience stronger winds? Did the failure model miss something about the attachment?

You work through the evidence before changing the model. Then you test whether the correction improves predictions across other homes and storms.

Model review / Roof damageExample
01
Reconcile the inspection, imagery and coded claim.
02
Check the hazard estimate and structural assumptions.
03
Build and benchmark the correction.
04
Reassess affected homes. Monitor the next event.
An illustration of the work, not a production result.
03 / Your authority

Own the model.
Answer for the results.

You own the risk decisions.

Build data sources and vulnerability models. Set house-level prices and selection rules within the company’s risk appetite. Resolve underwriting exceptions and define how the next similar case should be handled.

You make resilience measurable.

Identify which improvements reduce a home’s vulnerability, by how much and what each is worth in premium. Specify the evidence needed to verify the work and reprice the home.

You work with specialists.

Engineering builds and maintains the policy system. Insurance Operations runs it and coordinates partners. Our claims advisor sets the claims standard. You define the property and component data needed to test the model.

The data and models you will build

Machine-vision pipelines, structured inspection records, registry and permit integrations, and claim files coded by component. Structural simulations turn verified house characteristics and wind or hail fields into estimates of damage and loss. You calibrate against engineering science, vendor reference models and observed outcomes.

How we know it is working

Weekly reconciliation of predicted and actual losses, and expected loss against premium. Before storms, run the model against forecasts and share the exposed homes with Operations. After storms, investigate discrepancies, benchmark changes and monitor drift. Maintain a consistent property record across agents, inspectors, the carrier, claims administrator and reinsurer.

04 / What you bring

Quantitative rigor.
Accountability for risk.

You already bring

  • Models you built that informed decisions with real money at stake.
  • Production coding ability and strong quantitative foundations.
  • Experience testing predictions against outcomes and explaining uncertainty.
  • The ability to work across modeling approaches and learn unfamiliar techniques.
Where you might be coming from

Catastrophe reinsurance, insurance-linked securities, a reinsurance or catastrophe group at a multi-strategy fund such as D. E. Shaw or Two Sigma, mortgage credit, weather or energy derivatives, or catastrophe modeling on the buy side. The common thread is ownership of slow, fat-tailed risk and the data behind the decisions.

05 / The company

Close to leadership.
Close to the consequences.

Althea insures homes built or retrofitted to resist wind and hail, pricing each on its verified physical vulnerability. We are seed-funded, with a flat team, and are Launching Early 2027.

Jordan Breighner

President · Your manager

Capital, underwriting and modeling.

Geoffroy Bablon

Chief Executive

Technology, distribution and strategy.

Marc Ivins

Head of Sales

Agents and the referral channel.

Engineering team

System engineering

The policy system and the house record.

Insurance Operations

Also hiring

Policy system operations, partners, compliance and storm response.

You report directly to the President and own a critical system as an individual contributor. Your work shapes the risk we take and the incentives we offer homeowners.

06 / The process

Real work.
A clear process.

Two conversations, including a working discussion with Jordan. We will share the exercise and evaluation criteria in advance.

Step one

A conversation with Jordan

Talk through a poorly designed risk-pricing system. Where does it go wrong, what are the consequences and how would you improve it?

Step two

One home, an underwriting approach

Using a house photo we provide, walk us through how you would assess its wind and hail risk. Separate what you can observe from what you would need to verify. Lay out the data, model and assumptions you would use to estimate loss and inform a price, and explain how you would test them. We are evaluating your approach, not asking for a premium from a photograph.

Step three

A decision

We will discuss next steps and explain equity in full before any offer.

Start a conversation

Put your judgment
to work.

Get in touch about the role and the risk you have owned.

Apply for this role

Build a model that learns from every home we insure and every storm it lives through.

Your decisions determine which risks we take, how we price them and what we can offer a homeowner for making their house stronger.