Senior Data Scientist

Possible Finance•Seattle, WA

5d•$175,720 - $191,000•Hybrid

About The Position

Possible is a mission-driven fintech company helping everyday Americans build financial health through access to fair, affordable credit. Our data team sits at the center of a growing company, building the models, metrics, and experimentation infrastructure that powers how Possible makes decisions, measures performance, and operates with rigor at scale. We are seeking a Senior Data Scientist to work at the intersection of payments performance, optimization, and fraud, owning the analytical systems that govern how reliably money moves between Possible and our customers. This role expands our capacity to treat repayment as a designed experience rather than a back-end process, building payments that work with the rhythm of our customers' financial lives.

Requirements

Depth in the data science fundamentals and payments domain knowledge.
Experience with modeling, production monitoring, and experiment design.
In-depth understanding of payment rails (ACH, RTP, card, and interchange) and payment behavior.
Hands-on production ML development: built a model, deployed it, watched it drift, and retrained it, using tooling like XGBoost and MLflow or their equivalents.
Strong Python and SQL.
Comfort with large datasets in a distributed environment such as PySpark on Databricks.
Experimentation and causal inference skills.
Judgment to know which method a question calls for.
Feature engineering instincts for transactional data.
High bar for own work: understand data before drawing conclusions.
Willingness to find flaws in analysis.

Nice To Haves

Direct fraud modeling experience.
Background in collections, recovery, or lending operations in a regulated space.

Responsibilities

Own the data science behind how money moves at Possible.
Define and build the payments-health scorecard the company runs on, along with the monitoring that surfaces anomalies at the channel and experiment level within days.
Own how we time payments, sharpening how we identify a customer's pay cycle and designing retry strategies that work with it rather than against it, so that more payments clear on the first attempt: fewer failed-payment fees for customers, better recovery for the business.
Redefine how Possible understands payments fraud, building recurring reporting on the patterns that matter and developing a model that scores the risk of a new payment method or a payment that may not clear.
Work in Python, SQL, and PySpark on Databricks, with Datadog for monitoring and standard MLOps tooling for deployment.
Partner with Engineering, Product, and Risk to develop the payment strategy as an input to the engineering roadmap.
Act with ownership—defining what healthy payments means here, not waiting for a spec.
Take a scientific approach: rapid experimentation, intellectual honesty, and a willingness to change your mind when the data says so.
Optimize for customers ending up better off, not just for dollars collected.