Work · 6 projects · 2023 – 2026
Architecture case studies, the platforms I run, and what I built along the way. Every project starts from a real pain point and ends in production.
Design
Frame the need, choose the architecture.
Build
Platforms, pipelines and products, as code.
Run
Observe, document, stay until it runs.
Three systems told from the need to the result, and more below. Each one names the constraint, the decision and what it changed.
Medical AI training · 2025 – 2026
My first GPU platform, on Kubeflow, was barely used. I asked researchers why, moved to Slurm, and built a cluster that puts every GPU to work.
All GPUsshared through one queue, not only the fastest
Medical AI hackathon, Lyon · 2026
1 week to design, build and put the platform in production
Alfa-Safety · 2024 – 2025
2 TB
of logs a week, in one place
Smaller projects, school work and experiments. The code is on GitHub and GitLab.
