Freelancer profile translated to English.

Description

With 9 years of experience in data science and machine learning, I have developed a comprehensive mastery of the data lifecycle: from collection to production deployment, monitoring, and model retraining.

My approach combines scientific rigor, technical expertise, and an understanding of business challenges, with a particular focus on industrialization: delivering solutions that are robust in production, auditable, and maintainable by teams.

💼 Types of missions

Machine Learning Development & Engineering

I support companies in designing, implementing, and industrializing AI solutions tailored to their needs.

Design, training, and deployment of models (ML, Deep Learning)
Reproducible data and training pipelines (Data Engineering, MLOps)
Production deployment, observability, and governance (experiment tracking, drift monitoring, retraining)
Performance optimization and integration into existing products

Support & Skill Development

I help teams gain autonomy on data and AI projects to ensure the sustainability and internal mastery of solutions.

Tailored training (ML, NLP, LLMs, Data Science, MLOps)
Practical workshops and technical mentoring
Skill transfer up to production deployment

🧠 Areas of Specialization

In-depth expertise in the main fields of data science and ML:

Classical ML & Deep Learning: regression, classification, clustering, neural networks
NLP & LLMs: language processing, text generation, RAG
Computer Vision: image detection and recognition
Statistics: analysis, estimation, hypothesis testing
MLOps & Industrialization: reproducible pipelines, experiment tracking, deployment, monitoring
Ecosystem: Python, Scikit-learn, XGBoost, TensorFlow, PyTorch, LangChain, SQL, FastAPI, MLflow, Prefect, DVC, Evidently, Docker

Industry field of expertise

Languages

French
Native or bilingual
English
Fluent

Workplace preferences

Can work on-site

Paris (up to 50km), Lyon (up to 10km), Strasbourg (up to 10km), Nantes (up to 10km), Bordeaux (up to 10km)

Institut National de la Statistique, des Etudes Economiques et Démographiques (INSEED)
Senior Data Scientist / ML Engineer
PUBLIC SECTOR
September 2025 - Today (11 months)
Ndjamena, Chad
Design and industrialization of an end-to-end social targeting platform (PMT)

*Context:improving the targeting of poor households from a social registry using survey data, with strong requirements for sovereignty, auditability, and deployment in a constrained environment.

*Achievements:complete self-hosted MLOps architecture (FastAPI, MLflow, Prefect, Evidently, Prometheus/Grafana, Docker, DVC, PostgreSQL); reproducible data pipeline with survey weights and cluster validation; linear/ensemble/neural network benchmark evaluated on targeting metrics; stable variable selection (stability selection Lasso, permutation, SHAP) for short questionnaires; batch scoring, API, and offline artifact; SHAP explanation codes; governed retraining with human validation.

*Training and transfer:design and delivery of a curriculum (targeting statistics, regularization and variable selection, SHAP interpretability, neural networks, MLOps production deployment) using notebooks and materials; engineering mentoring (Git/GitHub, code review, reproducibility); transferable guides ensuring team autonomy after the mission.

*Measured impact:questionnaire reduced from 31 to 22 variables without performance loss; gradient boosting achieving 71% correct targeting of poor households in never-surveyed areas, 4.3 points higher (95% CI [3.3; 5.3], repeated cluster cross-validation) than the official regression on the same questionnaire; honest evaluation protocol also highlighting the optimism bias of the historical system.

*Skills:ML on tabular data, MLOps, industrialization, observability and drift, interpretability, model governance, sensitive data, skill transfer.
**Keywords**: MLOps · XGBoost · scikit-learn · MLflow · SHAP · FastAPI · Docker · Prefect · Evidently · DVC · social targeting · training
MLflow XGBoost SHAP Docker Scikit-learn
Datascientest / OMNES EDUCATION
Python, Data Science & Machine Learning Trainer
EDUCATION AND E-LEARNING
September 2024 - September 2025 (1 year)
Paris, France
Trainer for students at OMNES EDUCATION, I guided their learning of Python, data science, and machine learning through practical, project-oriented pedagogy.

*Python Programming:conducted practical workshops (basics, OOP, data manipulation) with key libraries — Pandas, NumPy, Scikit-learn, TensorFlow

*End-to-end Data Projects:supervised projects from collection to delivery — cleaning, transformation, and visualization using Python and SQL

*Modeling:assisted in the design and evaluation of Machine Learning and Deep Learning models

*Learning by Doing:solved real-world case studies and developed students' technical autonomy
Python Data Visualisation Machine Learning Deep Learning SQL