Lead Data Scientist

Remotefull timeRemoteseniorH1B: no

Paretocaptiveservicesllc The Lead Data Scientist at ParetoHealth leads applied AI and data science workstreams to improve risk selection, pricing accuracy, and operational efficiency in underwriting. The role includes end-to-end model development, deployment, monitoring, and contributing to AI product development and governance.

Responsibilities

Own end-to-end delivery and measurable outcomes for assigned predictive underwriting and pricing workstreams, from problem framing through deployment, monitoring, and continuous improvement, including:
Frame evidence-based recommendations and workstream trade-offs across model quality, risk, cost, scalability, aligning partners on delivery and measurable outcomes
Defining and documenting the Analytics-Ready Dataset, data-quality, and reusable feature requirements needed for assigned workstreams across claims, pharmacy, utilization, financial, underwriting, and external data
Applying rigorous point-in-time development and out-of-time validation approaches for claims maturity, seasonality, leakage, stability, and uncertainty
Developing, comparing, and challenging predictive models, distributions, and hybrid rule/model approaches based on evidence and operating constraints
Balance near-term delivery with disciplined exploration of emerging methods, including deep learning or GenAI solutions, that can materially improve accuracy, scalability, decision quality, or operating efficiency
Translating model needs into detailed feature requirements; partnering with business teams to identify additional signals and with Legal to secure approvals

Qualification

8+ years in data scienceAdvanced Python and SQLStrong business acumen and judgment

Required

Bachelor's or master's degree in Statistics, Data Science, Computer Science, Mathematics, Engineering, or a related quantitative field; an advanced degree is a plus
8+ years in data science, machine learning, statistics, actuarial analytics, including substantial work with healthcare, pharmacy, insurance risk, or sensitive longitudinal data and a proven record of independently owning models or analytical workstreams end to end
Advanced Python and SQL, with experience in scikit-learn, XGBoost, GBMs, or comparable frameworks. Experience using PySpark or comparable distributed-computing tools to work with structured and unstructured data at scale is preferred; PyTorch experience is a plus
Deep expertise in Supervised and Unsupervised Machine Learning methods, explainability, statistical distributions, rare-event and high-cost modeling, calibration, and optimization