Lead Risk Analytics Data Scientist

Extend is revolutionizing the post-purchase experience for retailers and their customers through AI-driven solutions. The Lead Risk Analytics Data Scientist will play a crucial role in risk monitoring and collaborate with cross-functional teams to maintain a healthy loss ratio while designing premium structures for new merchant categories.

Responsibilities:

• Price deals no one has priced before. Every time Extend enters a new merchant category, such as a new product type, a new vertical, or a partnership being negotiated right now, there's no existing model to hand you. You'll design the premium structure from first principles, figure out how to validate it when the data is thin, and own it through launch. The work is genuinely unsolved, and the decisions are yours to make
• Turn a monitoring function into a decision engine. We know what our loss ratios are. What leadership needs is to know what to do about them: which claims initiatives to pursue, which premiums to adjust, which programs to renegotiate. You'll build the tooling that makes that call possible, and you'll be the person making the recommendation
• Own the risk narrative with senior leadership. When the loss ratio moves, you're the one who explains why and what comes next, to an audience that includes the C-suite. If you've wanted more visibility than your current role gives you, this one puts you in that room
• Build the data foundation for Risk Analytics — and make it yours. We have dbt and Snowflake. What we don't have is a fully built Risk Analytics layer sitting on top of it. You'll design the data models, set the standards, and shape how this function works for the next several years. If you've ever inherited an architecture you wished you could rebuild from scratch, this is the rare opportunity to build it right the first time
• Actually use AI to change how the team works. Not as a buzzword, but as a working practice. We use Claude heavily. For this role, that means finding the places in the risk analytics workflow where a manual step shouldn't be manual, automating it, and raising the floor on what the team can produce. You'll have latitude to experiment and the expectation that you will