À propos de Quentin
Machine Learning Engineer — From Research to Production
Engineering student at Télécom Paris (IP Paris), with hands-on R&D experience at Mila / NeuroPoly (Montréal) and Synspective (Tokyo). I have built end-to-end ML pipelines that actually ship: optimized models, clean Python packages, and production-ready deployment stacks.
I have worked on a wide range of problems — medical image segmentation on multi-site MRI/CT datasets, semantic segmentation of SAR satellite imagery, and real-time signal processing — always with a focus on delivering something usable, not just a proof of concept.
My technical stack spans model training & fine-tuning (YOLO, Transformers, nnUNet, MONAI), inference optimization (ONNX, pruning, quantization), MLOps tooling (Docker, SLURM, Weights & Biases), and data pipelines (Python, SQL, NumPy, Pandas).
I have also worked on quantitative and statistical modeling as part of my applied mathematics curriculum, and I bring that analytical rigor to every project.
Primarily available for remote missions.
Français
Bilingue ou natif
Anglais
Capacité professionnelle complète
Espagnol
Capacité professionnelle limitée
En télétravail uniquement
Travaille majoritairement à distance
Expériences
- Mila – Quebec AI Institute & NeuroPolyMachine Learning Research Internmars 2026 - Aujourd'hui (3 mois)Montreal, QC, Canada• • Designed and open-sourced sc-crop (github.com/ivadomed/sc-crop), a Python package/CLI for automatic spinal cord detection and cropping in MRI/CT volumes, cutting segmentation inference from ∼2 min to ∼10–30 s• • Trained a 2.5D YOLO detector (Ultralytics) on multi-site, multi-contrast datasets (T1, T2, MP2RAGE, DWI, CT), achieving contrast-agnostic detection robust to distribution shift• • Built deployment stack: ONNX export + ONNX Runtime inference (authored nnunet-onnx), end-to-end detect–crop–infer–restore pipeline integrated with nnUNet• • Shipped production tooling: automated QC reports (CSV/JSON), 10+ versioned model releases, pyproject packaging, integration into the Spinal Cord Toolbox (SCT)
- SynspectiveMachine Learning R&D Internseptembre 2025 - février 2026 (5 mois)Tokyo, Japan• • Developed a modified SegFormer (Vision Transformer) for semantic segmentation of ocean-surface patterns in SAR satellite imagery, using Hugging Face Transformers• • Applied model compression (pruning, quantization) for on-board embedded deployment on a JAXA SAR satellite; containerized with Docker• • Contributed to design and evaluation of vessel-detection algorithms (object detection on SAR imagery)
- BDS T´ el´ ecom ParisCo-Organizerjanvier 2024 - janvier 2025 (1 an)Palaiseau, France• • Co-organized a 550+ participant relay marathon with 60+ volunteers: course design, logistics, safety, timekeeping, communication strategy
Recommandations
Soyez le premier à recommander Quentin
Contribuez à la réussite de ce freelance en partageant votre expérience de collaboration avec lui.
Ces profils de freelance correspondent également à vos critères
Agatha Frydrych
Backend Java Software Engineer
4.7
(3)
2
Baptiste Duhen
Fullstack developer
4.6
(4)
5
Amed Hamou
Senior Lead Developer
4
(2)
7
Audrey Champion
Web developer
4.3
(3)
4
Formations
- MSc-level Engineering Degree– Institut Polytechnique de Paris2027MSc-level Engineering Degree
- Intensive Mathematics & PhysicsLyc´ ee Henri-IV – Classes Pr´ eparatoires MPSI/MP2023Intensive Mathematics & Physics