Machine learning engineer, PhD. Computer vision for healthcare.
PhD at Inria Nancy and CHRU Nancy on detecting intracranial aneurysms in MR angiography.
I work on 3D detection and segmentation in medical images, on how clinicians review and override model output, and on what it takes to move a model from training to regulatory clearance and production monitoring: V&V, non-inferiority study design, PCCP, QMS, drift.
- DeepAneDet · DeepAnePose · DeepAneSeg — PhD pipeline: detection, 3D pose, segmentation.
- slicer-aneurysm-review — 3D Slicer extension for reviewing AI-detected aneurysm candidates: verdicts, rejection reasons, sign-off, provenance.
- slicer-weak-annotation — weak annotation as spheres, two clicks per lesion, no segmentation.
- classify-justify — a surface-defect classifier explained with 17 attribution methods, and whether those explanations can be trusted.
- CamemBERTQA — French extractive question answering: CamemBERT fine-tuned on FQuAD.
- travel-planner — a whole trip (route, itinerary, transport, budget, carbon) generated in the browser, no backend.
- AI–radiologist collaboration — Journal of Neuroradiology, 2025 · paper
- Automated aneurysm detection — IJCARS, 2024 · paper
- Anchor-free 3D pose estimation — MICCAI, 2023 (STAR award) · paper
- 3D segmentation under class imbalance — MICCAI DALI workshop, 2021 · paper
- PhD thesis, Université de Lorraine, 2024 · manuscript