About Soufiane
I design and operationalize machine learning-based systems and their evaluation frameworks — built through years of research and operational work, including at ECMWF, where I assessed the capabilities, physical consistency, and real-world skill of state-of-the-art AI weather forecast models.
With a Ph.D. in Climate Science and 6+ years of hands-on experience, I work across the full stack of data-intensive science: from building robust pipelines (NetCDF, GRIB, Zarr, Python, xarray, HPC) to developing verification tools used daily by operational forecast teams, to synthesizing results for technical reports and public-facing blogs. My PhD specialized in causal discovery methods applied to modes of climate variability, and I was acknowledged as a graphic developer in the IPCC Sixth Assessment Report (AR6).
I can help you with:
- ML model evaluation & benchmark design — defining metrics, designing diagnostics, assessing model skill and failure modes rigorously
- ML development & causal analysis — building prediction systems and applying causal discovery methods to understand dependencies in complex datasets
- Scientific data pipelines — reproducible, scalable processing workflows for large geospatial and time-series datasets in Python (xarray, pandas, numpy, scikit-learn)
- Energy-sector applications — wind and power ramp detection, forecast verification tailored to renewable energy use cases, and weather-to-energy diagnostics
- Forecast verification & validation — model comparison, uncertainty quantification, and performance monitoring
- Climate & weather data analysis — reanalysis, observational, and simulation datasets for research or product use
- Data visualization — scientific figures for publications, dashboards, and reports; IPCC AR6 contributor as graphic developer
- Technical writing & communication — translating complex results into clear insights for any audience
- AI/LLM tooling — tutorials, onboarding, and practical integration of agentic coding tools
English
Fluent
French
Fluent
Arabic
Fluent
Spanish
Basic
German
Conversational
Can work on-site
Bonn (up to 50km), Berlin (up to 50km), Hamburg (up to 50km), Cologne (up to 50km), Munich (up to 50km)
Experience
- ECMWF - European Center for Medium-range Weather ForecastsScientistENVIRONMENTALAugust 2024 - Today (1 year and 10 months)Bonn, Germany● Developed and implemented advanced evaluation frameworks for machine learning-based weather forecasts, with a focus on physical consistency diagnostics and process-based verification● Designed and operationalised verification tools for wind diagnostics, extreme-event analysis, and high-impact case studies● Built evaluation tools adopted by daily forecast analysts, improving routine model assessment workflows● Led the integration and evaluation of wind farm data within the Weather2x project, enabling user-oriented verification of forecast skill● Developed frameworks for wind and power ramp detection, including tailored evaluation metrics for energy-sector applications● Contributed to quarterly forecast evaluation reports, synthesising model performance, added value, and user-relevant impacts● Authored technical blog posts on the ECMWF's media center, communicating advances, strengths, and limitations in machine learning forecasts
- MARUM - Center for Marine Environmental Sciences, University of BremenPostdoctoral ResearcherENVIRONMENTALMarch 2024 - June 2024 (4 months)Bremen, Germany
- Diagnosis and Metrics for climate model evaluation
- Investigating air-sea interactions in high-resolution Earth System models
- Applying causal discovery techniques to study interaction in the southern ocean
- German Aerospace Center, Institute of Atmospheric Physics and Universität BremenPh.D. ResearcherENVIRONMENTALMarch 2020 - February 2024 (4 years)Bremen, Germany
- Investigate the causal relationships within the climate system and the effect of climate change
- Process, produce, and visualize extensive datasets to derive meaningful insights about the past, current, and future state of the climate system
- Evaluate climate model simulations in reproducing observed causal relationships
- Taught three Python lectures for geoscience postgraduate students
- Published extensive research in top-level peer-reviewed journal
- Presented results at the largest geoscience conference in Europe
- Implemented routines for the reproduction of IPCC AR6 figures using ESMValTool
Recommendations
Be the first to recommend Soufiane
Help this freelancer shine by sharing your experience working together.
These freelancer profiles also match your criteria
Agatha Frydrych
Backend Java Software Engineer
4.7
(3)
2
Baptiste Duhen
Fullstack developer
4.6
(4)
5
Amed Hamou
Senior Lead Developer
4
(2)
7
Audrey Champion
Web developer
4.3
(3)
4
Education
- Ph.D. in Climate ScienceUniversity of Bremen | German Aerospace Center (DLR)2024Causal model evaluation of teleconnections between major modes of climate variability in CMIP6 models
- M.Sc. in Environmental PhysicsUniversity of Bremen2020