Senior Data Scientist (66_2026.3)

Affinity Solutions (Affinity) is the leading consumer purchase insights company. We provide a complete view of U.S. and U.K. consumer spending, across and between brands, via exclusive access to fully permissioned data from over 100 million consumers. Our proprietary AI technology, Comet™, transforms these purchase signals into actionable insights for business and marketing leaders to drive optimal outcomes and build lasting customer relationships. Visit www.affinitysolutions.com to discover how we’re shaping the future of consumer purchase insights.

About Your Role:

Join Affinity’s Data Science team as a Senior Data Scientist, where you will apply advanced analytics, machine learning, and AI techniques to solve complex business challenges using one of the industry's richest consumer purchase datasets. In this role, you will partner with cross-functional teams to develop innovative models, generate actionable insights, and help shape the future of consumer purchase intelligence through cutting-edge data science solutions.

The Data Science team at Affinity Solutions builds the statistical and machine learning capabilities that turn raw credit card transactions into an AI-ready source of truth for consumer spending behavior — the models that resolve messy transaction strings into canonical brands and categories, the predictive models that turn spend history into forward-looking signals, and the methodology that measures campaign effects defensibly. Increasingly this work will be consumed by models and agents rather than by analysts, which raises the bar on correctness, robustness and privacy.

In this role, you will serve as the technical owner for the company’s Predictive Intelligence area. Your initial focus is our flagship spend-prediction system: taking ownership of the existing models that predict how much a card or individual will spend at a specific merchant over the next 30 days, the evaluation framework that measures their quality, the feature pipelines that feed them, and the weekly production scoring run that serves their output. From that foundation, the roadmap extends to longer-horizon brand and category forecasting, brand/ticker performance signals, behavioral embeddings, and other predictive and statistical intelligence — models and signals served as first-class, governed capabilities to our customers, our products, and the AI agents that will increasingly consume our data.

You will work with other Data Scientists and Engineers to take these capabilities from research to production at scale — designing models against billions of transactions, building the feature and backtesting infrastructure they depend on, and publishing their outputs as versioned, queryable attributes that any downstream consumer can trust.

Your Responsibilities

Immediate Focus: Owning Spend Prediction in Production

• Take ownership of the existing spend-prediction models, their evaluation framework, feature pipelines, and weekly production scoring — keeping them healthy, well-understood, and continuously improving.

• Design reproducible experiments with clear, upfront acceptance criteria for proposed model changes, and rigorously distinguish genuine improvements from noise, seasonality, or selection effects before rolling a change forward.

• When a metric regresses or an experiment result looks off, diagnose whether the cause is the model or an upstream data issue — late-arriving transactions, a schema or taxonomy change, a shift in the identity-spine join rate — before proposing a fix.

• Track and improve model quality against Snowflake compute cost per scoring cycle, treating credits spent as a first-class metric alongside accuracy, while meeting the scoring run’s required runtime.

• Build and refine the feature and label pipelines that turn the source/portfolio/card/transaction hierarchy into reusable, versioned features and model-derived attributes available to any downstream consumer.

• Develop metrics and quality measurement frameworks assessing quality, coverage, stability, and drift, and build the infrastructure to run these evaluations at scale

Roadmap: Extending Predictive Intelligence

• Engage in R&D to propose, explore, train, test and implement new predictive and statistical models beyond spend prediction — longer-horizon category and brand forecasting, brand/ticker performance models, and propensity and churn scores.

• Develop card- and individual-level behavioral embeddings and other representations that let customers build their own models on our data, trained where appropriate with differentially private techniques such as DP-SGD.

• Apply privacy-preserving modeling techniques — aggregation thresholds, perturbation-aware modeling, and cleanroom-compatible design — across both the current spend-prediction system and future models.

• Mine large consumer datasets in the cloud environment to identify new opportunities for ML.

• Direct AI coding agents effectively on modeling and engineering work. This codebase relies heavily on them. Maintain the documentation and context they depend on to work well, and critically review their output before it reaches production.

• Mentor junior team members in the areas of ML development.

• Serve as an ML expert within the company and provide ML support to clients and stakeholders.

• Communicate the methodologies and the results to management, clients, and other non-technical stakeholders.

Your Qualifications

• Extensive experience with leading production ML projects end-to-end, with emphasis on quality and scalability.

• Deep knowledge of the fundamentals of Machine Learning and Statistics. Proven ability to conceptualize business problems and craft sound and practical ML solutions.

• Strong experience with supervised learning on large-scale tabular and behavioral data — gradient-boosted trees, regularized regression, and neural approaches — including regression on sparse, zero-inflated, heavy-tailed targets such as consumer spend.

• Experience with time-series, panel, and forecasting problems.

• Proven experience designing reproducible model evaluation and backtesting frameworks — with clear acceptance criteria, fluency in calibration, ranking and uplift metrics and drift detection, and the judgment to separate genuine improvements from noise, seasonality, or selection effects.

• Comfortable diagnosing whether a model regression stems from the model itself or from an upstream data issue (e.g., delivery lag, schema drift, taxonomy changes, or join-rate shifts) and communicating that distinction clearly.

• Experience setting up infrastructure to support the ML development lifecycle, including reasoning about compute cost as a first-class constraint — evaluating quality-per-compute-credit tradeoffs and meeting runtime budgets at production scale.

• Strong software engineering and data engineering experience. Solid knowledge of Python and SQL. Experience writing production quality code.

• Experience working with cloud data warehouses such as Amazon Redshift, Snowflake, and data lakes such as Amazon S3.

• Extensive, hands-on experience directing AI coding agents on real production codebases — comfortable delegating substantial modeling and engineering work to them, maintaining the documentation and context they need to work well, and critically reviewing their output rather than accepting it at face value.

• Entrepreneurial, highly self-motivated, and collaborative, with keen attention to detail, the ability to learn quickly, and the ability to effectively prioritize and execute tasks in a demanding environment.

• Ability to communicate complex technical concepts and model results to management, clients, and other non-technical stakeholders through verbal, written, and presentation-based communication.

• Advanced degree in Statistics/Mathematics/Computer Science/Economics or other fields that provide advanced training in data modeling and analytics, and 5+ years of industry experience.

Preferred Qualifications

• Experience working with financial data, especially card transaction data.

• Experience or exposure to large consumer and/or demographic data sets.

• Experience with representation learning and embeddings for user, sequence, or behavioral data.

• Experience with privacy-preserving machine learning — differential privacy and DP-SGD, k-anonymity and aggregation thresholds, data cleanrooms such as AWS Clean Rooms, Snowflake, BigQuery, or Databricks.

• Experience with semantic or metric layers, feature stores, or otherwise publishing model outputs as governed, versioned data products.

• Exposure to LLMs and AI agents as consumers of model outputs through typed APIs and protocols such as MCP.

Salary Range: $185,000 – $200,000

Office Hours: 9:00 AM to 5:30 PM

Benefits for full-time employees of Affinity Solutions begin on the first of the month following your date of hire with a generous employer contribution for medical, dental, and vision. In addition to company paid holidays, wellness time off, other wellness benefits, and employee discounts, you will also get employer paid life insurance and have the option to enroll into an employer-matched 401K Plan. We strongly encourage work/life balance by providing unlimited vacation days, available starting 90 days from your hire date as a team member.