Freelancer profile translated to English.
Back to original languageAbout Bastien
Normalien (ENS Paris-Saclay), I specialize in Deep Learning applied to Physics (Graduated with a Master's in Applied Physics and Applied Mathematics).
Through the projects I have undertaken, I am capable of:
- preparing and cleaning complex databases,
- designing and training custom deep learning models (UNet, autoencoders, PINNs, ...)
- Fine-tuning existing models,
- rigorously validating performance (cross-validation, error analysis),
- writing a structured and interpretable report, providing decision support.
Due to my studies at the interface of physics/mathematics and AI, I am able to handle complex physical problems that a pure AI profile would not be able to handle adequately. I am proficient in Python, Numpy, Pytorch, JAX, Optuna, Weight And Biases tools. I also master C++.
I have had the opportunity to work on topics that allowed me to handle all the key stages of a Deep Learning project: database creation, model creation, hyperparameter optimization, results analysis, report writing.
French
Native or bilingual
English
Fluent
Remote only
Primarily works remotely
Experience
- Ecole Normale Supérieure Paris-SaclaySupervised Research WorkTECHSeptember 2024 - May 2025 (8 months)Gif-sur-Yvette, FranceTwo-semester research work during my Master 1 at Ecole Normale Supérieure Paris-Saclay. This work consisted of predicting damage and fatigue processes in crystalline structures by implementing a Unet-type convolutional neural network.
- Equipo de investigación Universidad de ZaragozaA Framework for prediction of cars drag coefficient and generation of optimized car shapes.TECHMay 2025 - July 2025 (3 months)Saragosse, SpainEnd of Master 1 university internship. Carried out at the Aragon Institute of Engineering Research (I3A), Zaragoza under the supervision of Mr Elias Cueto. During this internship, I worked on setting up an autoencoder-type neural network that allowed both the prediction of a vehicle's drag coefficient and the reconstruction of shapes by learning a signed distance. These two branches allow the generation of new optimized shapes according to their drag coefficient by interpolating in the latent space. The network was also geometrically informed, allowing for better reconstructions of vehicles during model inference.
Recommendations
Be the first to recommend Bastien
Help this freelancer shine by sharing your experience working together.
These freelancer profiles also match your criteria
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
Education
- Preparatory ClassesLycée Louis Lachenal2023Classes
- Master 1 Applied MathematicsUniversité Paris-Saclay2026Master d'Analyse, Modélisation et Simulation. Contient des cours d'analyse des équations aux dérivées partielles, Proabilités avancées, Statistiques, Optimisation.
Certifications
- IELTSBritish Council2024