Within Ipsos, the Creative Excellence helps clients understand what makes advertising effective across TV, digital, social and other media channels. A core strategic solution is Creative Spark AI, which is an AI-enabled capability to predict and explain ad performance, at scale and globally. This evolution will also include creation of new products utilizing the framework to take advantage of current blue ocean space in the market.
Requirements
- Computer Vision, GenAI and NLP
- Strong proficiency in Python and the main data & ML libraries (e.g. pandas, NumPy, scikit-learn, plus optionally TensorFlow / PyTorch / CatBoost / XGBoost).
- Good working knowledge of SQL and experience querying large analytical datasets (e.g. in BigQuery or similar cloud warehouses).
- Demonstrated understanding of core ML concepts: Feature engineering, regularization, model selection, cross-validation, Evaluation metrics for regression / classification, Bias, overfitting, drift, and robustness issues.
- Experience with NLP or Computer Vision applied to creatives (scripts, storyboards, video / image / audio).
- Experience with Cloud platforms, ideally Google Cloud Platform (GCP).
- Experience with Experiment tracking and MLOps tools (e.g. MLflow, model registries, CI/CD for ML).
Responsibilities
- Design, engineer, and test new model variants from survey, coded, or digital data sources, to improve prediction accuracy and explainability of ad performance across distinct ad environments and verticals.
- Integrate new features into experimental models and quantify their impact on prediction accuracy, robustness, and interpretability, and summaries the uplift (or lack thereof) for CRE senior management decision-making.
- Document feature definitions, derivation logic, and performance impact for replicability
- Design and execute experiments and benchmarks comparing different feature sets, algorithms, or model configurations (e.g. classical ML, deep learning, NLP / CV approaches).
- Use appropriate evaluation metrics (e.g. accuracy, AUC, RMSE, calibration, stability across segments) and validation schemes (cross-validation, hold-out, time-based splits) to ensure robust conclusions.
- Maintain clear experiment logs and documentation (notebooks, reports, dashboards) so results can be reviewed, reproduced, and reused by CRE and GADS teams.
- Contribute to continuous improvement of modelling best practices for Creative|Spark AI.
Other
- Master’s degree (or equivalent) in Data Science, Statistics, Applied Mathematics, Computer Science, Econometrics, or a related quantitative field.
- 7-10 years of professional experience as a Data Scientist in applied machine learning.
- Hands-on experience building and evaluating supervised learning models (regression / classification) in real-world use cases.
- Experience in product management or technical lead roles is a plus
- Experience in at least one of the following: Marketing, advertising, media, or market research, or Predictive modelling on survey, panel, or customer behavior data.
