SAID OBAKRIM
RESEARCH DATA SCIENTIST · STATISTICIAN
Location
Lausanne, Switzerland
Email
saidobak@gmail.com
GitHub
github.com/sobakrim
Homepage
wp.unil.ch/gaia/team/said-obakrim
Summary
Research data scientist and statistician with experience in academic research and scientific software. I design and implement statistical and machine-learning workflows, develop open-source tools, and communicate results across academic and applied settings. My work combines scientific modeling, software development, and collaboration around real research questions.
Experience
FNS Senior Researcher
University of Lausanne · Switzerland
- Conduct methodological projects in environmental and climate data science, from research design and workflow construction to implementation, validation, and documentation.
- Contribute to applied mandates on climate-risk management and natural-hazard assessment with ECAB, ECCE, and MOB.
- Supervise a PhD student and contribute to research communication through papers, technical reports, workshops, and presentations.
Postdoctoral Researcher – Weather Emulator
INRAE · France
- Designed and implemented
MSTWeatherGen, an open-source R package for multivariate space-time simulation.
- Led calibration, validation, and methodological documentation of the framework for epidemiology and wildfire-risk studies.
- The package is used in INRAE in epidemiological surveillance and environmental risk applications.
PhD – Statistics, Machine Learning and Environmental Data
University of Rennes 1 & Ifremer · France
- Developed statistical and deep-learning models for environmental time series, forecasting, and prediction in coastal climate applications.
- Published peer-reviewed articles and presented results at international conferences.
- Taught statistics and machine learning in Python and R, and supervised MSc-level work.
Selected Projects
mlq4st – Machine-learning Quantile Regression for Space-Time Processes
- Developed a python package for conditional distribution modeling and simulation of spatio-temporal processes using machine learning and latent Gaussian models.
- GitHub: github.com/sobakrim/mlq4st
MSTWeatherGen – Multivariate Space-Time Weather Generator
- Built an open-source R package for simulating multiple weather variables jointly in space and time for environmental and risk-related applications.
- Used in INRAE in epidemiological surveillance and environmental risk applications.
- Website: sobakrim.github.io/MSTWeatherGen
Technical Skills
Programming & Tools
Python (JAX, scikit-learn, TensorFlow/Keras),
R, Git, Linux, Jupyter, scientific software development.
R, Git, Linux, Jupyter, scientific software development.
Analytics & Methods
Statistical modeling · Machine learning · Deep learning · Extreme value modeling ·
Geostatistics · Stochastic simulation.
Workflow & Collaboration
Reproducible workflows · Documentation · Applied research collaboration ·
Technical reporting · Workshops and presentations.
Languages
Amazigh (native), Arabic, French, English.
Education
PhD in Statistics & Machine Learning
University of Rennes 1 · France
Thesis on statistical downscaling and climate-related modeling in coastal environments,
combining statistical methods and deep learning for environmental prediction.
MSc in Statistics & Econometrics
Mohammed V University · Morocco
BSc in Mathematics
Ibn Zohr University · Morocco