Senior Data Scientist - Predictive Audience Modeling Lead

Titan OS Partnership; Barcelona; On-site; EUR 90,000 - 140,000; Full time

Benefits offered by this job

Competitive pay

Private health insurance

Diverse international team

Opportunity to innovate in CTV

Job summary

Titan Operating System S.L. (Titan OS) in Barcelona seeks a Senior Data Scientist to lead predictive audience models for Connected TV advertising, transforming seed and observed data into robust population-level predictions.

This is a deeply technical role focused on statistics, sampling theory and machine learning, with emphasis on representativeness, selection bias, population extrapolation and uncertainty.

Qualifications

• 8+ years of professional experience in Data Science, Applied Statistics, Quantitative Research, Econometrics or a closely related field.
• Deep statistical expertise in sampling theory, estimation, inference, probability, uncertainty quantification and experimental design.
• Strong sample-to-population experience using post-stratification, raking, calibration/propensity weighting, hierarchical models, or MRP.
• Hands-on ML lookalike experience using seed audiences, CRM/customer samples or known converters with propensity, classification, similarity or positive‑unlabelled methods.
• Advanced ML & feature engineering skills across GLMs, XGBoost/LightGBM, ensembles, embeddings and large-scale behavioural datasets.
• Strong model validation skills including lift/gains, precision@K/recall@K, probability calibration, A/B testing, stability and drift monitoring.
• Advanced Python & SQL skills with scikit-learn, pandas/NumPy and experience with large-scale cloud data platforms and production ML/MLOps are highly desirable.
• Strong communication and quantitative foundations with a BS/MS in a relevant quantitative field; a PhD is advantageous but not required.

Responsibilities

• Build models that expand seed and sample audiences into high-value cohorts across the wider CTV population.
• Apply weighting, calibration, MRP and hierarchical methods to extrapolate reliably from imperfect or non-representative samples.
• Build propensity, classification, similarity, ranking and positive-unlabelled models to identify users with high affinity to a seed audience.
• Create robust behavioural and contextual features from large-scale event data while preventing target leakage and instability.
• Diagnose selection bias, coverage gaps and covariate shift, and communicate uncertainty in audience estimates.
• Measure lift, precision/recall, calibration and stability using holdouts, temporal/geographic tests and observed campaign outcomes.
• Move successful models into scalable pipelines with versioning, drift detection, monitoring and retraining.
• Partner with Product, Commercial and Engineering teams while setting rigorous modelling standards across the Data team.

Skills

Data Science

Statistics

Econometrics

Quantitative Research

Python

SQL

Machine Learning

MRP

A/B Testing

Uncertainty Quantification

Education

BS/MS in quantitative field

Tools

scikit-learn

pandas

NumPy

MLOps

XGBoost

LightGBM

Job description

Titan Operating System S.L. (Titan OS) in Barcelona seeks a Senior Data Scientist to lead predictive audience models for Connected TV advertising, transforming seed and observed data into robust population-level predictions.

This is a deeply technical role focused on statistics, sampling theory and machine learning, with emphasis on representativeness, selection bias, population extrapolation and uncertainty.