Data Scientist

Project overview

Stedin is looking for a Project Coordinator (Gas Grid Driven). (FREELANCER POSSIBLE) Indicative hourly rate is €95,- all-in, excluding VAT.

General

Within the Data Science Forecasting team, you will work on reliable short-term predictions of grid load. The assignment specifically focuses on implementing the OpenSTEF model for all LS transformers. OpenSTEF is the grid management wide open-source forecasting package for short-term (<7 days) predictions of grid load. Stedin uses these predictions for daily operations, particularly for capacity management and grid safety analyses. By the end of Q2 2027, the quality of all MS transformer predictions (23,000 assets) must be substantively improved (ACM deadline).

Goal and Content of the Assignment

Fully operational OpenSTEF implementation for the 23,000 LS transformer points with implemented features for quality improvement. Together with data scientists, machine learning engineers, the Product Owner, and stakeholders, you will determine which improvements yield the most forecast gain. You will research, implement, and validate these improvements, with explicit attention to production readiness, explainability, and reusability.

Skillset/Profile:

• Analysis of the current forecast quality and selection of locations, horizons, and situations with the greatest improvement potential.
• Backtesting and benchmarking of OpenSTEF methods against the existing reference, with appropriate quality metrics and segmentation.
• Improvements for difficult patterns, such as flatlines, peaks, missing values, unpredictable assets, and limited historical data.
• Experiments with model selection, ensembles, calibration, and confidence intervals.
• Exploration and validation of foundation model capabilities as soon as they are applicable and can be responsibly deployed.
• Translation of successful experiments into maintainable Python code, tests, configuration, and reproducible pipelines.
• Improvement of monitoring on forecast quality, data quality, and model behavior.
• Contribution to OpenSTEF, for example via issues, documentation, reviews, or pull requests, when an improvement is also valuable for the community.
• Knowledge transfer to the team and clear documentation of choices, results, and next steps.
• You can quickly test hypotheses, but also know what is needed to reliably land a model in a critical production chain.
• Completed university education in data science, econometrics, mathematics, statistics, electrical engineering, computer science, or a comparable field.
• Demonstrable experience with time-series forecasting and machine learning in a professional environment.
• Very good command of Python, including testing, code reviews, and version control.
• Experience with backtesting, experiment design, model comparison, and forecast metrics.
• Experience with production environments, MLflow/MLOps, and preferably Databricks and Azure.
• Familiarity with probabilistic forecasting, calibration, ensembles, or foundation models is a strong plus.
• Experience with open-source development or contributions to a technical community is a plus.
• You communicate clearly with both technical colleagues and business stakeholders and transparently document choices.
• You work independently, actively seek collaboration, and demonstrably transfer knowledge.

Project details

• Start: 2026-11-30T23:00:00
• Location: Rotterdam Zuid-Holland, NL
• Languages: Dutch