I’m an independent consultant in data science and machine learning engineering specializing in time series problems, with over 10 years of experience designing and delivering production-ready machine learning systems. I enjoy breaking down complex, time-dependent problems into simple, interpretable solutions and implementing them with clean, modular code. I strongly believe in open-source software and actively contribute to the Python ML ecosystem.

• I specialize in time series problems such as forecasting, anomaly detection, motif discovery, and classification, with applications in both industrial and business settings.
• I lead end-to-end data science projects built on Kedro, focusing on structure, reproducibility, maintainability, and smooth production deployment.
• I’m the author of kedro-dagster, which brings asset-based orchestration to Kedro using Dagster, and I regularly contribute to the Kedro and scikit-learn ecosystems.
• I have worked on several packages that extend or integrate with the scikit-learn API, though most remain closed source.
• My academic and technical background spans engineering, applied mathematics, physics, and high-performance computing, which strongly informs how I approach modeling, scalability, and system design.

• kedro-dagster
  A plugin for orchestrating Kedro pipelines using Dagster, a modern, asset-oriented orchestrator.

• giotto-tda
  Created during my postdoc at EPFL, this is an open-source Topological Data Analysis library for feature engineering and unsupervised learning, built on top of scikit-learn.

• metaLBM
  Developed during my PhD, this is a GPU-accelerated C++ simulation package for turbulence modeling using MPI, OpenMP, and CUDA.

You can reach me via LinkedIn or Mastodon.