Principal Data Scientist - Identity and Matching Models

🏢About Spokeo

Spokeo is a people intelligence platform that helps over 18 million monthly visitors reconnect with friends, reunite with families, and build trust in new relationships. Thousands of companies also trust Spokeo's 60 billion public records to improve customer research, help verify information, and prevent fraud.

Founded in 2006, Spokeo has built a dedicated, remote-first team with an average tenure of 6.9 years. Comparably has recognized it as a Best Company for Compensation, Employee Happiness, Perks and Benefits, Support for Women, Work-Life Balance, and CEO Leadership.

About This Opportunity

Every Spokeo search result depends on one decision: do these records describe the same person, and how sure are we? You will build the models that make that call, on messy data from many vendors and public sources, where there is no answer key to check against.

You will own a modeling problem from the first question to the production release, and you will see your work change what customers see. The Data Science team is small, so you work directly with the Director of Data Science and the product owners. If you like designing the test as much as building the model, and you want to mentor teammates as the team grows, this is the place to do it.

✅What You'll Do

• ✅Rough share of your time, subject to change.
• ✅35%: Build and improve the models that match records and score how confident we are in each result. You will choose the approach, train it, test it honestly on held-out data, and ship it.
• ✅20%: Design the labels and the measurement. Decide what to hand-check, build the evaluation sets, and estimate how often we are right when a full answer key does not exist. This includes using AI models to help label at scale, and checking them against human judgment.
• ✅15%: Take models to production on AWS, with ML engineering and DevOps. That covers the retrain schedule, the model registry, and a record of where every model and its data came from.
• ✅10%: Review each data build before it ships. Find what changed, catch regressions, and explain it in plain language to product and sales partners.
• ✅10%: Work with the Product Owner to turn findings into stories with clear targets, and defend the priority of the next research spike.
• ✅10%: Document your methods, review teammates' code, and take part in the DS/ML working group.
• ✅Tools. What you will work with: Python, SQL, Spark, AWS SageMaker, Databricks, and MLflow.
• ✅Challenges. Noisy and incomplete…