About Mathias Villerabel

Experience

AI Engineer

Equans | Paris, France | Jul. 2025 – Present

Engineering the full data pipeline from edge to model: Azure IoT Hub → Event Hubs/Kafka → Palantir Foundry streaming ingestion, with batch processing on Spark and Polars for historical backfill and feature stores
Building distributed feature engineering on Spark for predictive maintenance — rolling statistics, frequency-domain features, and cross-sensor correlations across fleets of industrial refrigeration units
Developing predictive maintenance models combining streaming IoT sensor data (temperature, pressure, vibration) with offline ML pipelines for anomaly detection and remaining useful life estimation
Implementing model serving infrastructure — online models for real-time alerting, offline models for long-horizon maintenance scheduling and fleet-level analytics
Operating end-to-end model monitoring: data drift detection, prediction quality tracking, automated retraining triggers, and SLA-based alerting
Owning the full ML lifecycle in production — from data engineering through training, deployment, A/B testing, and continuous monitoring

Data Scientist

Swiss Re | Zurich, Switzerland | Sep. 2023 – Jun. 2025

Built an IFRS 17 actuarial/financial simulation engine (cash flows, discounting, Contractual Service Margin) to benchmark EY outputs and ensure regulatory compliance
Automated end-to-end reconciliation (General Ledger ↔ actuarial models), cutting manual work from multi-day reviews to < 2 hours per run and improving auditability & reproducibility
Designed and optimized PySpark pipelines in Palantir Foundry, scaling to 10⁹+ accounting records per batch and integrating with downstream financial reporting systems

Data Scientist

Pernod Ricard | Paris, France | Jan. 2023 – Aug. 2023

Developed time-series forecasting and ML models (ARIMA, gradient boosting) and designed a custom PyTorch framework for scalable training/inference
Improved 3-month demand forecasts by 12% vs. Financial Analyst baseline

Research & Applied Scientist

Synspective | Tokyo, Japan | Dec. 2019 – Jan. 2023

Automated detection of new construction from SAR time series (InSAR coherence + intensity), improving urban growth monitoring
Delivered a cloud-based Earth observation platform with dynamic caching on Google Cloud, reducing data latency for SDG indicators
Built scalable object detection models for maritime trade (containers, cars, trucks, ships); awarded 2nd Prize at the NEDO Challenge
Applied ML at NASA, ESA & JAXA hackathons for wildfire monitoring and prediction

Machine Learning Intern

National Institute of Informatics | Tokyo, Japan | Mar. 2019 – Aug. 2019

Developed action detection models combining MobileNetV2 with segmentation/regression maps and LSTM for temporal recognition
Integrated deep learning into UAV systems for search & rescue; published in IEEE Access

Education

Master / Engineering Degree in Computer Science Sorbonne University (ex UPMC) | France Specialization: Machine Learning, Data Science, and Software Engineering

Technical Skills

ML / Deep Learning

PyTorch, TensorFlow/Keras, scikit-learn, XGBoost, LightGBM, CatBoost — time series forecasting (ARIMA, Prophet, LSTM, TCN, Transformers), computer vision (DeepLabV3, MobileNetV2, object detection, semantic segmentation), recommendation systems, SAR/InSAR analysis

Data Engineering

Azure IoT Hub, Kafka/Event Hubs, PySpark (10⁹+ records/batch), Polars, Pandas, NumPy, SQL, GeoPandas — streaming and batch pipelines on Palantir Foundry, Google Cloud, Azure, AWS SageMaker

Cloud / MLOps

Docker, MLflow, Hydra, Optuna, ONNX, GitHub Actions, pytest, pre-commit, SonarQube — experiment tracking, hyperparameter optimization, model monitoring, CI/CD

Languages

Python (expert), SQL (expert), C/C++, Java, Shell/Bash, MATLAB

Contact

Email: m.villerabel@gmail.com
LinkedIn: linkedin.com/in/mathias-villerabel-ba32b0146
GitHub: github.com/MrEleden

Download CV (PDF)