We are looking for a Senior Data Scientist to take ownership of a key part of our forecasting and marketing analytics capabilities.
The main challenge will be to evolve our current regression-based forecasting solution toward more advanced approaches, particularly Transformer-based architectures for time series forecasting.
This is not a role focused solely on implementing predefined models. We are looking for someone who can understand the business and data problem, define the modeling approach, experiment with alternatives, rigorously validate results, and ultimately help bring models into production.
You will have significant autonomy and technical ownership, working closely with the team while bringing your own judgment to modeling and architectural decisions.
Design, develop, train, and validate advanced time series forecasting models.
Evaluate Transformer-based forecasting architectures, including Chronos-2, as well as alternatives such as TFT, PatchTST, Informer, or other relevant approaches.
Build and run experiments using Python, PyTorch, NumPy, Pandas, and Scikit-learn.
Train and validate models using AWS SageMaker.
Design rigorous validation strategies for time series, including:
Compare different models and architectures, going beyond metrics to understand and explain why one approach performs better than another.
Translate forecasting results into actionable business insights and marketing decisions.
Contribute to Marketing Mix Modeling (MMM), attribution, and budget optimization initiatives.
Analyze channel contribution, incremental impact, saturation effects, and different investment scenarios.
Apply statistical inference, optimization, and model interpretability techniques to turn predictive outputs into business recommendations.
Clearly communicate technical decisions, assumptions, results, and trade-offs to both technical and business stakeholders.
Strong experience in Machine Learning and Data Science, with a particular focus on time series and forecasting.
Hands-on experience building and validating forecasting models in real-world scenarios.
Strong understanding of time-series validation methodologies and the challenges associated with evaluating models on future data.
Strong proficiency in Python.
Experience with PyTorch and the scientific Python ecosystem, including NumPy, Pandas, and Scikit-learn.
Experience working with AWS SageMaker for model training and experimentation.
Solid background in statistics, statistical inference, optimization, and model interpretability.
Ability to independently approach open-ended problems, from data exploration and methodology selection to model validation and actionable conclusions.
Strong analytical judgment and the ability to justify modeling and architectural decisions.
Ability to connect technical results with business outcomes.
Experience with Transformer architectures for time series forecasting, particularly Chronos-2.
Experience with other forecasting architectures such as Temporal Fusion Transformer (TFT), PatchTST, or Informer.
Knowledge of Bayesian Optimization or tools such as scikit-optimize.
Experience with Marketing Mix Modeling (MMM), marketing attribution, or causal modeling.
Understanding of concepts such as:
Experience with tools or frameworks such as Robyn, PyMC-Marketing, or LightweightMMM.
You will be able to take a relatively open-ended data problem and drive it from initial exploration to a validated solution.
Success in this role is not just about achieving better model performance. It is about choosing the right methodology, validating it correctly, explaining why it works, and translating the results into decisions that create measurable business value.
We're looking for someone who combines strong technical expertise, rigorous experimentation, and business thinking — and who is comfortable taking ownership of complex modeling challenges.
