Three strands come together in this practice. Analytical foundations from applied mathematics and statistics, with two peer-reviewed publications behind them. Hands-on machine learning and generative AI expertise. And the engineering discipline that takes both into production.
That combination was built in order. The work started in statistics and data analysis, moved into data science and predictive modelling, then into the engineering side: making models reproducible, deployable and maintainable, which led to MLOps, cloud and production environments. More recently it has extended to AI engineering, generative AI, RAG, agents and LLMOps.
Four-plus years of delivery across Luxembourg, Switzerland, France and Burkina Faso, mostly in healthcare, finance and technology, where an AI system has to be not just accurate but reliable, explainable and defensible. The point is never a model that stays in a notebook: it is turning an idea into something a team can operate and trust.
Research-grade rigor
Every model is evaluated honestly, with metrics chosen for the problem, not for the headline. Rooted in applied mathematics and peer-reviewed research.
Production is the goal
A model in a notebook is not a deliverable. The finish line is a reproducible, tested, deployed system a team can operate and trust.
Governance by design
In healthcare, finance and insurance, AI must be explainable, monitored and compliant. Governance is built into the architecture, not bolted on.
Business value over buzzwords
The work is judged by outcomes and the confidence it inspires, never by the length of a technology list.
Machine Learning & Deep Learning
Generative AI & LLMs
MLOps & Cloud
Data & Engineering
APIs & Applications
Governance & Foundations
2026-Present
AI / GenAI Consultant & MLOps Engineer
Sahel Politica · Bern, Switzerland
Turning nearly two decades of political and security risk expertise on West Africa into AI-enabled intelligence capabilities, from a source-grounded analyst assistant to an automated tender-intelligence platform.
- Domain-specific GenAI and RAG assistant connected to the in-house knowledge base
- Analysts retrieve, synthesize documents and produce source-grounded intelligence insights
- Tender intelligence platform covering public and private calls across Europe and Africa
- Requirement analysis and first-draft application packages, generated then refined by experts
- Human in the loop by design: the system never submits anything on its own
2025-2026
AI × MLOps Engineer
HALE-X · Luxembourg · Hybrid
Designing and deploying scalable AI within robust MLOps pipelines for a company using AI and digital-twin technology to turn multi-source health data into insight for hospitals, laboratories and clinicians.
- Built end-to-end MLOps pipelines for AI-driven healthcare applications
- Developed ingestion and preprocessing across multi-source medical data
- Trained, fine-tuned and evaluated predictive and generative models for clinical use cases
- Contributed to reliable deployment on cloud infrastructure
- Worked with clinicians and cross-functional teams on compliance, scalability and performance
2024-2025
AI Trainer · Prompt Engineer
DataAnnotation · Remote (USA)
Engineered and evaluated prompts and AI-generated code across data analytics, data science and statistics use cases.
- Designed and tested specialized prompts for data science and statistics problems
- Evaluated generated answers and validated generated code, identifying technical errors
- The work was judging whether a solution was actually correct, not merely plausible
- Improved workflow efficiency by roughly 25% through automation
2025
Machine Learning / MLOps Engineer
Freelance · Self-Employed · Geneva, Switzerland · Hybrid
Built a production-ready credit-default prediction system for a financial institution, on a heavily imbalanced dataset, from EDA to a serverless API on AWS with full CI/CD.
- Exploratory analysis and feature selection, then SMOTE to handle class imbalance
- XGBoost, Random Forest and Logistic Regression compared on metrics fit for imbalance
- Hyperparameters tuned with RandomizedSearchCV before model selection
- FastAPI service containerized with Docker, automated tests with pytest
- CI/CD with GitHub Actions, deployed on AWS Lambda behind a custom domain
2024
Data Scientist (Trainee)
Luxembourg Institute of Health · Strassen, Luxembourg
Estimated influenza cases from wastewater viral loads on the VIRALERT project, working with noisy and incomplete data, in collaboration with the Luxembourg Institute of Science and Technology.
- Authored the Statistical Analysis Plan that guided the pipeline
- Benchmarked statistical and machine learning approaches, including LSTM models
- Built interactive Plotly visualizations so the research team could read the results
- Contributed to the associated scientific publications
2020-2022
Data Analyst · Statistician
Centre MURAZ, Institut National de Santé Publique · Bobo-Dioulasso, Burkina Faso
Statistical analysis and scientific writing across public-health research programmes, where the quality of the analysis directly shaped how the findings could be interpreted.
- Statistical modelling across COVID-19, HIV, pesticide and fertility studies
- Supported the ANRS COV-13 EMuL-COVID-19 clinical research project
- Contributed to the POCAO West-Africa key-populations programme
- Co-authored a peer-reviewed study on pesticide exposure and infertility
Master's Degree in Applied Mathematics & Statistics (Data Science)
Université Rennes 2 · L'Institut Agro Rennes-Angers
Rennes, France
Co-accredited by Université de Rennes, Institut Agro, INSA and ENSAI. Statistical learning, machine/deep learning, time series, Bayesian modelling.
Bachelor's Degree in Applied Mathematics: Statistics & Computer Science
Université Nazi Boni
Bobo-Dioulasso, Burkina Faso
Have an AI system that needs to reach production?
Let's talk about the problem you're trying to solve and whether AI is the right tool for it. No buzzwords, just an honest technical read.