Freelancer profile translated to English.
Back to original languageAbout Christophe
Your data science projects failing? Interesting AI ideas but fragmented data and POCs that don't scale?
I am a Data Scientist & ML Engineer with 10 years of experience across the entire data value chain. I have supported clients in various sectors: finance, retail, logistics, ...
What I do for you:
🔹 Production-ready data pipelines (Python, GCP, DBT) — data science foundation for your ML & GenAI models
🔹 Entity resolution, master data management, and data quality — up to 90% reduction in duplicates
🔹 ML models in production (MLOps): scoring, predictive analytics, and operational optimization
🔹 AI agents and LLMs that transform your data into business actions (RAG, data retrieval)
🔹 Documentation and transfer of data expertise to your teams
My approach:
In data science, I first build a measurable baseline, then iterate. Clean infrastructure to facilitate maintenance and evolution. This is how we move from initial results to continuous improvements.
My latest results:
âś… Data quality: -90% duplicates via data science & entity resolution (EurosForDocs)
âś… ML data engineering: +95% restaurant matching rate between UberEats and Deliveroo (Clone / UberEats)
âś… ML forecasting & optimization: -20% food waste thanks to ML vs manual strategy (Foodles)
✅ Fraud: -20% financial losses on B2B fraud (€200k/year recovered) via ML scoring (Ellisphere)
My data stack:
- Python: Django, flask, pandas, polars, python-igraph, networkx, scrapy.
- Machine Learning: scikit-learn, LightGBM, XGBoost, fastai, MLFlow.
- SQL: BigQuery, PostgreSQL, DuckDB, Snowflake.
- Orchestration: Airflow, DBT, Dagster.
- Google Cloud Platform: Composer, Cloud Run, Cloud SQL, Compute Engine, Docker.
- BI & Visualization: Metabase, Looker.
đź“… Free data science audit. 30 min to identify your 3 quick wins: data quality, ML, GenAI.
French
Native or bilingual
English
Fluent
Can work on-site
Bordeaux (up to 10km), Paris (up to 10km), Toulouse (up to 10km)
Experience
- CloneData Scientist - AI Agent for RestaurateursRESTAURANTS AND FOOD SERVICEMarch 2024 - November 2025 (1 year and 8 months)Paris, France
- Multi-Agent AI on GCP via Telegram: < 10s latency, 95%+ success rate
- Deduplication of 1M restaurants with 95%+ matching via fuzzy matching + graph + ML
- Data pipeline for extracting KPIs from Deliveroo
🔎ContextClone supports 1000+ restaurateurs in managing multiple menus across platforms (UberEats, Deliveroo, ...) to increase their revenue.The goal is to provide a GenAI copilot for a single point of entry to actions and data available on the platforms.📊TasksCopilot Agent for Restaurateurs.Creation of an AI agent (Gemini, Google ADK) accessible via Telegram. Multi-agent architecture with an orchestrator and two specialized sub-agents:- Action agent to modify restaurant status (price, menu, ...),- Analytics agent interfaced with a data warehouse (Looker + BigQuery) for activity monitoring.Demonstrated performance: latency below 10 seconds, success rate (tool calls + correct parameters) above 95%.Entity Matching.I created a data science pipeline (python, polars, igraph) to link 1 million restaurant accounts from various platforms to physical restaurants (including dark kitchens) and provide a competitive analysis (BigQuery, DBT) accessible to the agent. +95% matching on specific datasets for edge cases.Performance Monitoring Pipeline.Data pipeline (airflow + BigQuery + DBT) for daily extraction of operational metrics (sales, promotions, costs) from partner restaurants' platform accounts.🛠Data & GenAI Stack-Data Science:pandas, polars, scikit-learn, python-igraph-Data Engineering:Airflow, DBT-ML Engineering:MLflow-Backend:Django, Scrapy, Telegram, Docker-Databases:BigQuery, PostgreSQL, Google Sheets, Parquet-Cloud:Google Cloud Platform (GCP)-GenAI/LLM:Gemini, Google ADK - Association EurosForDocsLead Data Scientist | Expert in Data Quality & Cost OptimizationCIVIC AND SOCIAL ORGANIZATIONSAugust 2023 - Today (2 years and 10 months)Paris, France
- 90% reduction of a database of health professionals through duplicate merging. → Trust in data established for citizens and journalists.
- Mentoring a team of 6 juniors.
🔎 Context:
EurosForDocs is an association that fights conflicts of interest in healthcare by making a public database of physician remuneration by pharmaceutical companies accessible.The project involves improving data quality, particularly unifying the numerous duplicates in the database.đź“Š Tasks:
- Reassignment of national identification numbers (RPPS) for health professionals via fuzzy matching with the official directory.
- Identification of duplicates of medical associations using NLP and machine learning.
- Creation of a unique entity grouping all duplicated accounts (entity matching).
- Minimization of infrastructure costs through the use of Python and DuckDB.
- Implementation of best development practices (GitLab CI/CD, peer review, consistent naming) within the junior team.
đź› Technical Stack:
- data science: pandas, python-igraph, scikit-learn, LightGBM, python
- data engineering: Dagster
- machine learning engineering: MLFlow
- SQL: DuckDB, parquet, Google Sheet
- data visualization: Metabase
- FoodlesSenior Data ScientistRESTAURANTS AND FOOD SERVICEApril 2021 - October 2024 (3 years and 6 months)Clichy, France
- Reduction of 20% food waste
- Support for growth, from 50 to 300+ clients.
🔎 **Context**:Foodles is a player in corporate catering, offering daily delivery of fresh meals in connected refrigerators to enable teams to eat healthily at any time of day.The project involves implementing an operational optimization system to distribute50,000 meals weeklyto300+ clientswhile minimizing food waste and maximizing consumer satisfaction, a critical system integrated into the company's backend infrastructure.📊 **Tasks**:- Training an AI system for customer sales forecasting.
- Development of an optimal meal distribution solution among clients, limiting food waste and ensuring consumer choice. A critical system requiring a solution within a defined timeframe for daily operations.
- Support for business teams in adopting the results of data tools.
- Segmentation of diners and meals to personalize refrigerators based on consumption habits in each company, guiding operational optimization in its food variety choices.
đź› Technical Stack:
- Python: Django, CVXPY
- Machine Learning: scikit-learn, LightGBM, PyTorch, MLFlow
- SQL: PostgreSQL, Snowflake
- DBT
- Docker
Reviews
Recommendations
LM
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
- Doctor of Philosophy - PhDUniversité Paris-Saclay2017Doctor of Philosophy - PhD