IW
Ibrahima Wane
Portrait of Ibrahima Wane

Ibrahima Wane

Data scientist with 2 years of hands-on banking experience, building ML systems end to end — from messy raw data to explainable models and deployed services. Currently at Société Générale; MIAGE master's graduate from Paris Dauphine.

Selected work

Project explorer

Filter by discipline to see how I combine analytics and engineering.

Accessibility Atlas — Île-de-France

2026

Mapping public-transit inequalities across the 1,297 communes of Île-de-France: a composite mobility score from open GTFS and INSEE data, used to identify 17 priority transit deserts.

  • Composite score blending walk distance, daily frequency, and modal diversity, with a log-transform to keep the scale sensitive at the low end
  • 17 priority deserts isolated by combining population (≥3,000) with a sub-median score — 10 of them in Seine-et-Marne, zero in the Transilien-served Yvelines
PythonGeoPandasShapelySciPyFoliumGTFS

Credit Risk Scoring — Freddie Mac

2026

End-to-end default risk model on Freddie Mac mortgage loans: reproducible Kedro pipeline, SHAP explainability, a Streamlit app, and a RAG documentation assistant.

  • Chose logistic regression (AUC 0.74) over XGBoost for regulatory interpretability
  • SHAP revealed the feature ranked #1 by importance actually hurt out-of-sample AUC
Pythonscikit-learnXGBoostSHAPKedroStreamlit

MLOps — Road Accident Severity

2026

MLOps pipeline classifying French road-accident severity (BAAC 2021) behind a containerized FastAPI service with inference, retraining, and monitoring endpoints.

  • Random Forest on ~54k accidents served via FastAPI + Docker Compose
  • Endpoints for prediction, on-demand retraining, health and metrics, with CI
PythonRandom ForestFastAPIDockerCI/CD

Bank Marketing Conversion Prediction

2024

Binary classification predicting term-deposit subscription from a bank marketing campaign, handling strong class imbalance end to end.

  • Compared logistic regression, random forest, SVC and KNN on imbalanced data
  • Applied SMOTE resampling after full EDA on categorical and numeric drivers
Pythonscikit-learnSMOTEPandas

Bayesian Inverse Reinforcement Learning

2024

Recovering reward functions from observed behavior in a gridworld MDP using Bayesian IRL with the PolicyWalk sampling algorithm.

  • Built a gridworld MDP, policy iteration, and an imperfect-tutor simulator
  • Inferred rewards via Bayesian PolicyWalk sampling over Q-values
PythonReinforcement LearningBayesianNumPy

Experience

Industry impact

Société Générale

Data Scientist Apprentice

Sep 2024 - Present

La Défense, France

  • Rebuilt the pre-screening model for professional prospects: estimates the grantable credit amount before relationship onboarding, cutting case-processing time and smoothing the client journey.
  • Brought explainability to the banking recommendation engine serving 7 million retail clients — SHAP analysis plus an automated dashboard advisors use to read each decision.
  • Segmented retail clients to sharpen commercial targeting and tailor offers by risk and behavioral profile.
PythonPySparkSQLDataiku

BMCI

Full Stack Developer Intern

May 2023 - Sep 2023

Nouakchott, Mauritania

  • Rebuilt the bank intranet with a modern architecture to improve the user experience.
  • Implemented secure file storage and user access management.
  • Delivered advanced search, filtering, and document viewing features.
Spring BootAngularMySQL

Université Paris Dauphine

Data Engineer Intern (Research Support)

Jan 2023 - Mar 2023

Paris, France

  • Automated SQL reports and data extractions for research teams.
  • Improved processes to integrate internal tools.
  • Updated user guides and provided support for adoption.
SQLDocumentationSupport

Education

Master MIAGE — Business Intelligence

Université Paris Dauphine · 2023 - 2025

Bachelor's Year 3 — MIAGE

Université Paris Dauphine · 2022 - 2023

  • Coursework: data analysis (regression, PCA, statistical tests), probability & statistics, linear programming (simplex, duality), relational databases & SQL.

Bachelor's — Mathematics & Economics

Université Claude Bernard Lyon 1 · 2019 - 2022

  • Coursework: data analysis (PCA, clustering, supervised classification, neural networks), advanced probability & statistics, econometrics, operations research.

Skills

Toolkit

Languages & Databases

PythonSQLJavaRC/C++PostgreSQLMySQLMongoDB

Machine Learning

pandasNumPyscikit-learnTensorFlowPyTorchXGBoostSHAPMatplotlib

MLOps & Engineering

FastAPIStreamlitKedroDockerGitHub ActionsJenkinsTerraformAzureBash

Data & BI

Apache SparkDataikuTableauPower BIJupyter

Get in touch

Let's turn data into decisions

I'm looking for a full-time role as a data scientist or data analyst. If you're hiring, I'd love to talk about how I can help your team ship models that actually reach production.

Journey

From Nouakchott to data science

Nouakchott2019 · Lyon2023 · Paris

I grew up in Nouakchott, Mauritania, where a fascination with mathematics took hold early. The numbers were the easy part; what stuck was how they could be used to describe and decide. That curiosity drifted toward statistics, then toward data — anywhere a problem could be framed clearly and answered with evidence.

On August 22, 2019, at 18, I packed my bags and left for France. The first stop was Lyon, at Université Lyon 1 Claude Bernard, where I began a Bachelor's in Mathematics & Economics. The mix of formal proof and real-world models was a good start, but it pushed me toward a more applied path.

So I reoriented. I moved to Paris and joined the MIAGE program at Université Paris Dauphine — a curriculum built around the meeting point of computer science, decision-making, and data. It turned out to be the right fit. I earned my Master 2 there, and the projects on this site are the work that came out of that period: explainable models, MLOps services, real datasets, real trade-offs.

Alongside my studies, I led the conference pole of Paris Dauphine Afrique, the university's African student association. We put together three conferences — on journalism in Africa, on African sovereignty, and on the war in the DRC — and a round table on African weddings between tradition and modernity. Bringing those conversations to a French university audience was its own kind of work, different from data but just as worth doing.

Outside of school, I travel when I can — a few European trips (Lucerne, Milan, Barcelona, Amsterdam) and the cliffs of Étretat closer to home. Dakar stayed with me the longest: a way of staying close to West Africa, where my story started. Photography goes with all of this — paying attention to what's in front of me long enough to want to keep it.