Staff Applied Scientist, LTV Modeling

Faire•New York City, NY

54d•Hybrid

About The Position

As a Staff Applied Scientist on the Discovery team, you'll own how Faire measures and optimizes the long-term value of a discovery impression — one of the highest-leverage open problems on our marketplace. Our rankers today optimize for order conversion, helping retailers find brands and products they love. But we know our ranking algorithms can do more: helping retailers find not just products they love, but brands they can build long-lasting, successful partnerships with. Reordering is one clear signal of this — successful brand-retailer relationships compound into substantial reorder volume over time — but not every discovery order evolves into a lasting partnership. Identifying the ones that will compound, and helping them grow, matters enormously for our community. This is a rare opportunity to define a measurement problem from first principles. You'll build the LTV framework, design the experiments that validate it, and turn the result into a shared signal that all discovery algorithms can act on.

Requirements

5+ years applying ML and statistical modeling to real business problems, shipping to production.
Deep causal inference expertise — quasi-experimental methods, rigorous confounder control, and healthy skepticism of analytical results.
Strong experimentation design skills, especially long-horizon experiments — surrogate/proxy metrics and variance reduction for sparse, delayed outcomes.
Baseline knowledge of search and recommendation systems on e-commerce or marketplace platforms.
Strong statistical analysis and data engineering skills — SQL/ETL and data transformation at scale.
An excitement and willingness to learn new tools and techniques.
Excellent communication skills and the ability to work in a highly cross-functional team.

Nice To Haves

PhD in CS, Stats, Economics, OR, or a related STEM field.
LTV / lifetime-value optimization on a two-sided marketplace, e-commerce platform, or other recommendation systems.
Deep learning, machine learning, or learning-to-rank techniques.

Responsibilities

Own how we measure and optimize the long-term value of a discovery impression — how it contributes to the discovery of new brands retailers might love, and how it strengthens existing promising relationships so they compound.
Create the initial LTV framework: form and prioritize hypotheses about what drives long-term relationship value and the key short- vs. long-term tradeoffs, with assumptions made explicit and testable, and lay out the experimentation roadmap to validate and refine it.
Lead the implementation of v0 of the LTV model into a long-running ranking experiment, setting north star metrics as well as guardrails to maximize organizational learning, with a defined readout cadence and course-correction plan.
Deliver the long-term surrogate metric — a near-term readout predictive of long-term value — accounting for confounding factors and inherent uncertainties in measurement and marketplace dynamics.
Own the LTV model tech stack and operating standards, continuously improving the capabilities and accuracy of the model as it becomes consumable across search, reorder, and ads.