A broader view of the work.

Paris-Dauphine finance and machine learning project

Loan approval scoring pipeline from experimentation to batch inference

Refactored an academic credit scoring study into a modular Python project with training, feature engineering, batch inference, business-facing risk bands, and Docker packaging.

Advanced unsupervised learning and audio representation project

Unsupervised audio pipeline with handcrafted features and Transformer embeddings

Built a full unsupervised audio clustering workflow on 769 environmental clips, then benchmarked classical acoustic descriptors against AST, wav2vec2, and HuBERT embeddings.

Paris-Dauphine analytics and data visualization project

Clustering, profiling, and dashboarding for marketing decisions

Performed unsupervised customer segmentation on 2,240 clients, compared clustering families, and packaged the results in a Shiny dashboard for business exploration.

Applied reinforcement learning project inspired by a medical control paper

Reinforcement learning for adaptive treatment policies under constraints

Implemented a reinforcement learning controller for chemotherapy dosing, simulated several patient profiles, and reproduced a robustness study over 15 synthetic patients.

Time series and actuarial applications project

ARIMA, GARCH, ARIMAX, VAR, and causal analysis across market and macro data

Built an end-to-end econometrics workflow on L'Oreal stock prices and French macroeconomic series, covering stationarity, forecasting, volatility modeling, and multivariate dynamics.

Applied mathematics and reinforcement learning project in C++

Reusable C++ framework for Value Iteration on discrete decision problems

Implemented generic MDP classes, a dynamic programming solver, tests, and a robot garbage collector example in modern C++.

Monte Carlo methods coursework in applied mathematics

Simulation, rare-event estimation, and variance reduction with visual explainers

Implemented five Monte Carlo methods with animated visualizations to explain inverse CDF sampling, accept-reject sampling, stratification, importance sampling, and control variates.

Applied machine learning extension of an academic GLM project

Model comparison platform for operational demand forecasting

Built a forecasting workflow for bike rental demand with six models, Box-Cox target transformation, model tracking, and deployment-ready packaging.

Biomedical signal processing project based on an INTERCON paper

Benchmark of DWT, PCA, and Kernel PCA on noisy cardiac signals

Reproduced and benchmarked ECG denoising methods on MIT-BIH data, comparing DWT, PCA, and Kernel PCA across several noise families.