Decision Scientist

The Decision Scientist will support the design, development, and deployment of Next Best Action (NBA) models that determine the optimal marketing action across channel, promotion, and creative. These models power decisioning across owned channels including email, SMS, and push, enabling personalized, data-driven customer engagement at scale.

This role sits at the intersection of marketing strategy and data science, and requires an ability to work collaboratively with marketing, technology, and merchandising partners.

• Design and build end-to-end Next Best Action (NBA) decision models that optimize marketing channel, promotion type, creative variant, and send timing at the individual customer level
• Support or contribute to reinforcement learning, multi-armed bandit, or contextual bandit frameworks as part of the NBA decisioning engine, with opportunity to grow expertise in this area
• Develop propensity models (purchase, churn, reactivation, category affinity) that serve as inputs to the NBA decisioning engine
• Build and maintain customer-level response models measuring incremental lift from marketing interventions across email, SMS, and push
• Collaborate with marketing technology teams to deploy models into real-time or near-real-time decisioning environments (e.g., via API or CDP integration)

• Design and analyze A/B and multivariate experiments to measure model performance and continuously refine decisioning logic
• Partner with campaign operations to translate model outputs into actionable audience segments, suppression lists, and treatment assignments
• Apply optimization approaches that balance short-term revenue goals with longer-term customer engagement and retention objectives
• Build holdout and incrementality testing infrastructure to ensure accurate measurement of model-driven lift

• Mine transactional, behavioral, and engagement data to engineer predictive features at the customer level
• Build and maintain customer feature stores supporting NBA model inputs: recency, frequency, category affinities, channel responsiveness, and promotional sensitivity
• Integrate third-party data sources (demographic overlays, loyalty data) to enrich model inputs and improve prediction accuracy
• Develop deep understanding of Belk customer segments by loyalty tier, shopping occasion, and FOB affinity to ensure models reflect behavioral nuance

• Write clean, well-documented code in Python and/or R for model development, feature engineering, and scoring workflows
• Build SQL-based data pipelines to extract, transform, and prepare modeling datasets from enterprise data platforms
• Establish model monitoring, drift detection, and retraining cadences to maintain model accuracy over time
• Document model methodology, assumptions, validation results, and performance benchmarks to support governance and reproducibility

• Partner with marketing strategists and CRM leads to define decisioning use cases and prioritize the model development roadmap
• Translate complex model outputs and findings into clear business narratives for non-technical marketing and business stakeholders
• Contribute ideas and best practices within the Decision Science function, and collaborate effectively across analytics and marketing teams

• Bachelor’s Degree in Statistics, Mathematics, Computer Science, Data Science, Economics, or related quantitative field required.

• 2-4 years applied data science, quantitative analytics, or related work; hands‑on experience with predictive modeling in an academic or professional setting required.
• 1-2 years building models in a retail, e‑commerce, marketing, or related business context preferred.
• Experience deploying models into production environments; familiarity with CDP platforms (e.g., Salesforce Marketing Cloud, Adobe, Braze) a strong plus.

• Expert‑level proficiency in Python and/or R for statistical modeling, machine learning, and data manipulation
• Deep knowledge of supervised and unsupervised ML algorithms: gradient…