Freelancer profile translated to English.
Back to original languageAbout Geneviève
I have been a data scientist for 2 years and previously worked for over 15 years as a researcher in fundamental (quantum) physics.
My positioning in data science is currently quite generalist, with particular expertise intabular data processingusing usualmachine learningmethods and in **natural language processing**.
With extensive experience in scientific research, I bring rigor, method, and clarity to the projects I support.
⚙️ Stack: Python, SQL, GCP, Dbt, Power BI
🚀 Feel free to contact me to discuss your data projects or to learn more about my achievements.
French
Native or bilingual
English
Conversational
Can work on-site
Palaiseau (up to 30km)
Experience
- DoleadData ScientistDIGITAL AND ITAugust 2025 - September 2025 (1 month)In the Data & Science team (4/5 people) of the startup Dolead, specializing in digital marketing.Short mission (20 days) aimed at exploring the possibility of predicting the number of leads generated by Facebook advertising campaigns, based on (among other things) the parameters provided as input during the setup of these campaigns:● Exploratory data analysis (Pandas, Plotly)● Training and evaluation of supervised regression models (Scikit-learn)● Documentation and communication
- Office français de la biodiversitéData scientist (Fixed-term contract)ENVIRONMENTALSeptember 2024 - April 2025 (7 months)Vincennes, FranceIn the Data and Methodological Support Unit (~15 people)Development of a frugal AI tool for extracting structured data from PDFs:● Data analysis on public water and sanitation services (SISPEA)● Acquisition, analysis, and labeling of PDF reports on the price and quality of water and sanitation services (RPQS)● Development in Python of the NARVAL tool to extract SISPEA indicator values from RPQS PDF reports: a frugal AI approach combining light heuristic methods and small language models (SLM)● Evaluation on a corpus of 45 PDFs: definition and analysis of metrics, incremental improvement process of NARVAL after error analysis● Estimation of carbon impact (1g CO2eq per PDF)● Automation of the import of NARVAL's output file to SISPEA (notebook only)● Critical analysis of the developed solution: limited to non-scanned RPQS from "small" communities, 91% precision and 84% recall for the solution calling the SLM, 100% precision and 48% recall for the solution extracting only indicators from summary tables, ...● Discussion of prospects● Documentation and communicationPYTHON - PANDAS - HUGGING FACE TRANSFORMER - SQL - GIT--------------------------------------------Side project (<10% of time): cross-referencing Naïades data on water quality with Carthage data on water body mapping● Exploratory analysis of Naïades and Carthage databases● Creation of station graphs including all Naïades stations upstream and downstream of a given station● Some examples of physicochemical data visualization on these graphsPOSTGRESQL - POSTGIS - pgROUTING - PYTHON - PANDAS - GEOPANDAS - GIT
- DoleadData scientist (freelance)DIGITAL AND ITOctober 2023 - April 2024 (6 months)In the Data & Science team (4/5 people) of the startup Dolead specializing in digital marketing.Contribution to the development of a lead scoring pipeline:● Data cleaning and analysis (SQL, pandas, plotly)● Feature addition and selection● Training, selection, and calibration of supervised classification models (logistic regression, random forests, gradient boosting) using BigQuery ML + dbt● Use of results to feed Google/Meta bidding models and impact study● Documentation and communication
Reviews
Recommendations
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
- Data Scientist Bootcamp TrainingDataScientest.com (partnership Mines Paris Tech PSL)2022Formation bootcamp, parcours data scientist (mai à juillet 2022) 300h de 𝐜𝐨𝐮𝐫𝐬 sous la forme de notebooks dirigés et masterclasses: ● Programmation : fondamentaux Python, Numpy, Pandas ● Data visualisation : Matplotlib, Seaborn, Bokeh ● Machine Learning avec scikit-learn : classification, régression, clustering, réduction de dimensions, régularisation, text mining, séries temporelles (avec statsmodels) ● Deep Learning avec Keras et Tensorflow: réseaux de neurones denses, convolutifs, récurrents, auto-encodeurs, GANs, optimisation des architectures neuronales et des hyperparamètres, régularisation, transfer learning, introduction au reinforcement learning ● Introduction au Data Engineering: bases de données, SQL, PySpark 100h de 𝐩𝐫𝐨𝐣𝐞𝐭 𝐟𝐢𝐥 𝐫𝐨𝐮𝐠𝐞 en trinôme sur un sujet au choix (parmi 7): détection non supervisée de sons anormaux 👉 https://github.com/gefleury/datascientest_anomalous_sounds
- PhD in theoretical physicsUniversity Paris 62010Préparée au CEA Saclay d'octobre 2006 à octobre 2009, soutenue en janvier 2010. Simulations Monte Carlo Quantique des effets de l'interaction coulombienne sur la localisation d'Anderson dans le gaz 2D d'électrons.
Certifications
- Data ScientistDataScientest & Mines ParisTech PSL2022
- Microsoft Certified: Power BI Data Analyst AssociateMicrosoft