Freelancer profile translated to English.
Back to original languageAbout Sophie
Senior statistician with over 20 years of experience at INSEE in complex data analysis, with dual training in data science (recent master's) and social sciences (research master's). First co-author of a scientific paper (ERIC Lyon 2, LIRIS CNRS) presented at INSEE's JMS 2025 on deep learning graph analysis.
Producing numbers is easy.
Understanding what they actually measure is much less so.
My role: intervening when data exists but indicators are questionable, their limitations poorly identified, and decisions difficult. I help formulate analyses and models that are truly usable for decision-making.
Framing and reformulating statistical and data problems
(unclear objectives, questionable indicators, metrics disconnected from decision-making)
Robust statistical analyses
(descriptive, inferential, reasoned segmentation, indicators)
Data science and modeling adapted to the context
(when relevant, and only when it is)
Sociological analysis of data, uses, and markets
(understanding populations, collection biases, exclusion effects)
Qualitative contributions in social sciences
(semi-structured interviews, narrative analysis, data contextualization)
Targeted expertise and bibliographic research
to shed light on little-known markets, products, or populations
Philippe Manipoud (CEO, Alberture)
"[Sophie] quickly distinguished herself by her ability to deeply analyze complex data, drawing on her skills in sociology, statistics, and data science.
Her strength also lies in her ability to present her results clearly, structured, and concisely, making her deliverables immediately actionable.
Sophie brought real added value to our strategic thinking."
[...]
👉 All my recommendations are on linkedin: linkedin.com/in/sophie-perrin-statisticienne
French
Native or bilingual
English
Fluent
German
Conversational
Spanish
Fluent
Icelandic
Basic
Can work on-site
Lyon (up to 10km)
Experience
- AlbertureData Scientist R&D – Anomaly Detection on Graphs and Sociology of Banking InclusionTECHMarch 2025 - July 2025 (4 months)Givors, FranceR&D project in data science applied to human and social sciences, resulting in a scientific paper presented at the INSEE's Statistical Methodology Days (JMS) 2025, in collaboration with CNRS (LIRIS – Université Lyon 1) and the ERIC laboratory (Université Lyon 2). Supervised by Cyril Perosino (CIFRE PhD student within a deep tech startup) - project completed during my master's final internship.Methodological, technical, and interpretative contributions, as well as educational and commercial valorization of results to non-technical audiences.My Contributions👉Modeling and R&D:Contribution to the design of a graph anomaly detection model (GNN), using supervised and semi-supervised learningComparative study of hierarchical clustering methods in hyperbolic geometry (HypHC) and spectral clustering, to improve model performanceProposal of data from human and social sciences relevant for enriching model inputsConsideration of the regulatory framework for a model classified as "high risk" by the AI Act.👉Valorization, Pedagogy, and Communication:Co-authoring educational and commercial materials, translating complex mathematical concepts (fairness, explainability, clustering, graph anomaly detection) into clear and differentiating messages for non-technical stakeholdersDesigning, based on an economics thesis, a summary slideshow (65 slides), as well as a summary and an executive summary for communication and decision-making aid purposesOrganization and ToolsMultidisciplinary teamwork (data science, HSS, marketing)Development in Python, use of collaborative tools (Git, Jupyter, VS Code)
- inseeSenior Statistician – Public Data Analysis, Surveys, and Data QualityPUBLIC SECTOROctober 2003 - Today (22 years and 10 months)Lyon, FranceIntervention on methodology and statistical analysis issues, ensuring thedecision-making use of resultsand their **understanding by non-specialist audiences**.
Skills Used:
Statistical Analysis- Analysis of survey data, includingvery large volumes
- Evaluation ofdata quality(non-response, bias, structural effects, collection methods)
- Construction ofindicatorsandtypologiesfor decision support
Programming- Data processing inR, SQL, SAS
Knowledge Transfer and Valorization- Writingstatistical publicationsandanalysis notes
- Presenting results toinstitutional**, **professional**, and **media audiences
- Explanation ofmethods used in constrained contexts(e.g., health crisis)
Examples of Achievements:
National analysis of theeffects of interview collectionon the quality of population census dataWritingregional statistical studies(Auvergne-Rhône-Alpes), publicly disseminated, andpresentingthese analysesat press conferencesCreation of a typology of non-responding establishments in the tourism survey, allowing foreffective targeting of follow-upsAnalysis, **combining qualitative sociological interviews and statistical analysis**, of the causes of persistent non-response from certain establishments in the monthly hotel attendance survey, to improve data collectionStudy on non-salaried workers:innovativeuse of a **complex administrative database**, **inter-regional technical coordination**, creation of shared SAS scripts and a variable dictionary - Compagnons du Devoir ; éducation nationaleMathematics TeacherEDUCATION AND E-LEARNINGSeptember 1999 - June 2002 (2 years and 10 months)Lyon, FranceItaughtvery diverse audiences:middle schoolers**, **high schoolers**, and **young adults in manual apprenticeships(evening classes after physically demanding workdays for them).These experiences taught me tostructure clear explanations**, to **effectively simplify complex notions**, and to **maintain attentioneven in contexts not conducive to concentration.
Recommendations
Be the first to recommend Sophie
Help this freelancer shine by sharing your experience working together.
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
- Master 2 MIASHS (Mathematics and Computer Science Applied to Human and Social Sciences)Université Lyon 22025Mention bien (après accès direct sans avoir fait de master 1) Contenu : data science - machine learning - deep learning - analyse de réseaux Communication "Combiner clustering hiérarchique et spectral pour la détection d'anomalies dans des graphes : application aux réseaux sociaux" présentée aux Journées de la Méthodologie Statistique (JMS) de l'INSEE 2025
- Research Master 2 - Social and Cultural AnthropologyUniversité Lyon 22010Mention très bien (après accès direct au niveau Licence 3) Mémoire de recherche "L'inceste : consistance du silence - France, 20 et 21e siècles" lauréat du 2e prix scientifique de la CNAF (Caisse Nationale des Allocations Familiales)