Data Scientist III

Hagerty

1d•Hybrid

About The Position

As a Data Scientist III at Hagerty, you'll build the customer identity and personalization layer that powers how we understand and engage members across our subscription and property & casualty (P&C) insurance products. This is a hands-on, build-and-ship role on the Data Science team, working in close partnership with ML Ops, Data Engineering, and Marketing/Product. You'll help create a unified, resolved view of each member across our data ecosystem—spanning auto insurance policies, subscription memberships, and the broader automotive enthusiast community—and turn it into recommendation, personalization, and predictive models that deliver the right message at the right moment. The goal is a system where identity, relevance, and timing work together to make every member interaction feel personal—at scale.

Requirements

Hands-on experience designing, training, and deploying ML models in production.
Proficient in Python and modern ML frameworks such as scikit-learn and XGBoost.
Strong in SQL and comfortable with large, distributed data platforms (e.g., Snowflake, SQL Server, AWS RDS).
Experience with identity resolution and entity matching using deterministic and probabilistic techniques.
Experience building recommendation or personalization systems, including content-based and/or hybrid methods and cold-start strategies.
Experience developing predictive models for customer behavior (churn, propensity, next-best-action, or similar).
A practical understanding of real-time vs. batch serving and the latency considerations that shape model design.
Familiar with production-ML concepts—containerization, API-based serving, and orchestration—and able to collaborate with ML Ops and Engineering to ship.
Able to turn ambiguous objectives into clear, data-driven approaches and executable plans.
A clear communicator who can tailor technical explanations to different audiences.

Nice To Haves

A background in P&C insurance, subscription or membership businesses, or financial technology a plus.
Master's degree (or equivalent practical experience) in Data Science, Computer Science, Engineering, Mathematics, or a related quantitative field.
3+ years of hands-on machine learning and data science experience, including models deployed to production.
Direct experience with a Customer Data Platform (CDP) and activation/audience workflows.
Experience with graph modeling or knowledge graphs applied to customer or relationship data.
Familiarity with our production toolset, or close equivalents: Docker or Podman for containerization SageMaker Endpoints or FastAPI for model serving Metaflow or Airflow for workflow orchestration
Exposure to anomaly detection, embeddings, or feature stores supporting real-time use cases.
Experience owning a meaningful slice of the lifecycle—from research through deployment and monitoring—in partnership with ML Ops or platform teams.

Responsibilities

Build identity resolution across first-party and third-party data sources, stitching member, household, vehicle, and behavioral signals from auto insurance and subscription touchpoints into a coherent, usable view.
Develop matching systems that pair a strong deterministic foundation with probabilistic matching at scale, balancing precision, recall, and cost.
Partner with Data Engineering and the Customer Data Platform (CDP) team to land resolved identities and audiences into production pipelines and activation systems.
Help evolve the identity layer toward graph-based representations of members, vehicles, and policy/membership relationships.
Design, build, and evaluate recommendation and personalization models, including content-based and hybrid approaches, to surface next-best-product and content across our insurance and subscription offerings.
Develop cold-start strategies that deliver relevant experiences to new and low-engagement members.
Make deliberate trade-offs between real-time and batch serving, designing models and features with latency and freshness constraints in mind.
Build well-calibrated predictive models for member behavior across the P&C and subscription lifecycle—churn/retention, propensity to buy, and propensity to lapse or renew.
Develop next-best-action and journey-signal models that translate behavior into triggers the business can act on, supporting cross-sell and upsell across insurance and membership products.
Own full modeling workflows: exploratory analysis, feature engineering, model development, cross-validation, and performance monitoring.
Ship models as reliable production services in partnership with ML Ops, contributing to containerized deployments, automated testing, and monitoring.
Source and analyze features from Snowflake, SQL Server, and AWS RDS Postgres, and work with Data Engineering to promote proven features into scalable pipelines.
Contribute to the team's modeling standards through maintainable, well-documented, testable code.
Communicate methods, results, and trade-offs clearly to technical and non-technical partners.