Senior Data Scientist I, Monetization

Fetch•US - Remote,

About The Position

Fetch’s AI & Data organization brings together data engineering, analytics engineering, machine learning, experimentation, and data platforms to create business and customer impact. Practitioners work hands-on across production systems, analytical foundations, and intelligent features, collaborating with product, platform, and engineering partners while balancing quality, performance, scalability, and responsible use in complex, evolving problem spaces. As a Senior Data Scientist I on the Monetization team, you will independently own a defined audience or measurement product, primarily supporting Fetch’s third-party data products business and applying what you learn to first-party user monetization. You will work across the product lifecycle: building informed audiences, extending them beyond Fetch users, measuring off-platform outcomes through receipts, and using results to improve future audiences. This hands-on scientific and product role partners closely with Product, Engineering, Sales, and GTM. You will also engage directly with customers to explain product methods, value, and limitations. The role can be performed from a Fetch US office or remotely within the United States.

Requirements

At least 5 years of experience in data science, machine learning, or applied analytics, with demonstrated ownership of complex analytical problems in product-driven environments.
Experience owning an audience, measurement, or data product from an ambiguous question through delivery and a business decision.
Track record of regular collaboration with product, engineering, and commercial partners, and ability to explain customer needs and personal contributions clearly.
Experience creating useful datasets from fragmented, incomplete, or biased sources and demonstrating how their validity was assessed.
Strong population inference skills, including generalizing from an observed sample to a broader population and assessing who is missing.
Experience building propensity or audience models and validating real-world usefulness through population coverage, calibration, and measured outcomes—not only offline accuracy—and using results to improve subsequent iterations.
Hands-on causal inference experience, including designing defensible experiments or observational studies, explaining comparisons, assumptions, diagnostics, and limitations, and distinguishing attribution from incrementality.
Strong SQL, Python, applied statistics, and predictive modeling skills, with a habit of building reusable, reproducible assets.
Ability to explain scientific findings clearly to nontechnical and customer audiences.
Sound judgment about when to work independently and when to seek review because assumptions or scope exceed the evidence.

Nice To Haves

Applied experience in monetization data science, audience development, advertising measurement, market research, or consumer panels, especially when data work was tied to customer or commercial value.
Experience working with panel, survey, consumer research, or purchase data.
Background in political science, survey methodology, econometrics, demography, or another quantitative field involving strong datasets built with limited resources and explicit inferences about populations from sparse data.

Responsibilities

Independently own a defined audience or measurement product from problem framing through delivery.
Translate advertiser goals and purchase behavior into audience definitions, labels, and predictive models.
Build and validate statistical expansion beyond Fetch users, assessing quality, calibration, and reach on held-out populations and distinguishing observed purchases from inferred propensity.
Combine receipt, partner, panel, survey, and contextual data into reliable datasets; account for linkage uncertainty, missingness, and coverage; and apply weighting or calibration when warranted.
Design credible studies connecting ad exposure to receipt-observed purchases, using experiments when feasible and justified observational methods otherwise.
Assess receipt coverage, exposure matching, selection bias, and uncertainty, while distinguishing attribution from incrementality.
Turn customer problems into scoped products with clear success criteria and launch recommendations, balancing audience quality, reach, advertiser value, user value, revenue, and cost.
Use campaign outcomes, holdouts, and fresh data to improve audiences and models and verify that improvements are real.
Partner with Sales and GTM to shape pilots and explain product value, methods, and limitations without overpromising precision or lift.
Build reproducible datasets, analyses, and validation checks, and partner with Product, Engineering, and ML engineers to move prototypes into production.
Document methods, share reusable code, and communicate learnings to the team.