This role requires the ability to work lawfully in the U.S. without employment-based immigration sponsorship, now or in the future.The Head of Data Science builds the models that turn a new AI-native behavioral intelligence platform into products customers buy, and proves to those customers that the products work. At the core of the platform, part of Spectrum Intelligence Ventures, is a proprietary transformer-based model that learns how households behave from network-scale signal and encodes that behavior as embeddings. Those embeddings are the foundation for audiences, insights, measurement, and data licenses sold to brands, agencies, and ad tech, media, and AI platforms. Over the next 18 to 24 months, this leader's job is to get that portfolio to market and make it stand up to buyer scrutiny: build the predictive models and products on top of the embedding layer, set the evidence standard buyers will test them against, and work alongside product and sales to scope pilots, answer technical buyers, and convert evaluations into contracts. In parallel, this leader owns the direction of the embedding model itself, working with the Head of Technology and external development partners, and is accountable for showing that it delivers performance conventional approaches cannot. The Head of Data Science builds and leads the data science team, sits on the platform leadership team, and reports directly to the Head of Intelligence Ventures.
HOW THE HEAD OF DATA SCIENCE MAKES AN IMPACT - Build and run the predictive models behind the platform's audience and data products, including in-market, life-stage, purchase propensity, category interest, lifestyle affinity, and behavioral velocity models scored daily at the household level across more than two dozen verticals. Own segment definitions, coverage, refresh cadence, and measured accuracy for every audience the platform sells, whether through Spectrum Reach, DSP data marketplaces, or partner platforms.
- Own the platform's Feature Store as a commercial product: named, typed, versioned, and documented features that serve audiences, insights, measurement, and partner models. Lead the outcome anchoring program that trains and calibrates models against licensed third-party outcome data (vehicle registrations, mortgage originations, card and SKU-level retail spend, travel bookings, and similar sources) so that propensity scores mean what they say, and automate feature creation as new signal sources come online.
- Lead the modeling behind custom and partner-specific audiences, including seed-based lookalike expansion, profile-based audience construction, and site- or CRM-seeded models that identify households entering a brand's purchase path before they reach the brand. Establish repeatable, clean-room-based workflows with transparent match rates, expansion multiples, and performance against the buyer's own baseline.
- Design and defend the platform's measurement methodology, including exposed-versus-control campaign lift using matched behavioral controls, test-and-holdout incrementality, and closed-loop validation against real-world outcomes. Define the journey, demand, and category metrics that power insights subscriptions and aggregate demand indices, built to withstand scrutiny from brand marketing science teams, agency analytics groups, and measurement partners.
- Lead the data science behind the platform's identity and viewing products: the adjudication and confidence calibration methods that verify partner household graphs against network ground truth, and the household viewing-state, trajectory, and reachability models that underpin viewing intelligence for programmers, agencies, and measurement companies.
- Direct the roadmap for the platform's proprietary transformer-based behavioral embedding model, a multi-entity architecture that encodes both a stable household profile and short-term intent, working with the Head of Technology and external development partners on architecture, training, and refresh. Hold the embedding layer to a commercial standard: measured lift over attribute-only and engineered-feature baselines in the audiences, lookalikes, measurement, and partner models it feeds.
- Establish the evaluation, MLOps, and responsible AI standards for every model the platform ships, including evaluation cards that state accuracy, calibration, and coverage in a form buyers can cite; automated release gates; monitoring and drift detection; bias and fairness testing; and privacy-preserving techniques such as audience floors, differential privacy, and adversarial testing, in partnership with Legal and Privacy.
- Serve as the senior data science voice with customers and partners, including brand analytics and insights leaders, agency data science teams, and DSPs, identity providers, measurement companies, and AI platforms. Scope pilots and evaluations with product, sales, and solutions teams, and translate methodology into clear commercial proof points.
- Build and lead a team of data scientists, ML engineers, and measurement scientists matched to the work ahead: strong in applied predictive modeling and measurement, with the deep learning depth to take greater ownership of the embedding layer over time. Set hiring priorities, operating cadence, and a culture of rigorous experimentation with a clear bias toward production deployment and revenue impact.
REQUIRED QUALIFICATIONS- Proven record leading data science for data, audience, identity, targeting, or measurement products in the advertising, media, or marketing ecosystem, where models are packaged, priced, sold, and renewed by external buyers
- Deep, hands-on expertise in predictive modeling on consumer-level data at scale, including propensity and response modeling, classification, gradient boosting, regression, calibration, segmentation and clustering, and lookalike modeling, with sound judgment about which method fits the commercial problem
- Strong grounding in measurement and causal inference, including incrementality testing, matched and synthetic controls, test-and-holdout design, and the practical limits of attribution and media mix approaches
- Working knowledge of identity and data collaboration: household and device graphs, onboarding and match rates, clean rooms (Snowflake, LiveRamp, AWS, or equivalent), identity namespaces such as RampID and UID2, and the use of licensed third-party data in modeling
- Command of the full data science lifecycle in production, from feature engineering through training, validation, deployment, monitoring, and iteration, at large dataset scale (billions, even trillions of records); experience designing feature stores and scoring pipelines that serve audiences and partner deliveries
- Hands-on proficiency with Python, SQL, and distributed data processing (Spark or equivalent); experience running ML workloads on Snowflake, Databricks, AWS, or equivalent cloud platforms
- Solid working knowledge of deep learning and representation learning, including embeddings, sequence models, and transformers, sufficient to direct an embedding roadmap with external partners, set its evaluation standards, and put embeddings to work as features in downstream models
- Familiarity with privacy-preserving analytics and consumer data regulation, including aggregation thresholds, differential privacy, and consent management; comfortable working alongside Legal and Privacy on product design
- Ability to explain methodology clearly to executives, sellers, and buyers, and to defend it under scrutiny from sophisticated technical counterparts in external partner and customer conversations
- Track record of building and growing high-performing data science teams, ideally from an early stage, and recruiting senior talent in competitive markets
- Curiosity about the application of AI to consumer intelligence, including how foundation models and AI agents can draw on behavioral context to make better decisions
REQUIRED EDUCATION- Bachelor's Degree in Computer Science, Statistics, Mathematics, Economics, or a related quantitative field
PREFERRED EDUCATION- Master's degree or PhD in Statistics, Machine Learning, Econometrics, Operations Research, or a related quantitative discipline; an advanced degree is valued but not required, and a record of applied commercial impact is weighted equally
REQUIRED RELATED WORK EXPERIENCE AND NUMBER OF YEARS- Experience leading applied data science or analytics teams that build data, audience, identity, or measurement products: 7 years
- Hands-on predictive modeling, machine learning, or measurement science on consumer-level data at scale: 10 years
PREFERRED RELATED WORK EXPERIENCE AND NUMBER OF YEARS- Senior data science leadership at a data, identity, audience, or measurement company in the advertising and media ecosystem (e.g., LiveRamp, Merkle, Epsilon, Acxiom, Experian, TransUnion, Nielsen, Circana, Comscore, The Trade Desk, or specialty analytics and measurement firms): 10 years
- Applied deep learning or representation learning (embeddings, sequence models) deployed in production on consumer behavioral data: 3 years
- Work with TV and video viewership data, cross-screen measurement, or identity graph quality evaluation: 3 years
WORKING CONDITIONS- In-office position preferably based in New York City
- Travel as required for partner engagements, executive meetings, and industry events
#LI-SD2
The base pay for this position generally is between
$263,200.00 and $393,800.00. The actual compensation offered will carefully consider a wide range of factors, including your skills, qualifications, experience, and location. We comply with local wage minimums and also, certain positions are eligible for additional forms of other incentive-based compensation such as bonuses.
