Über Julien
Bonjour !
Je suis Julien, Senior Data Scientist (10 ans d'expérience)
Diplômé de l'École Polytechnique, je combine un large bagage mathématique avec une maîtrise des architectures IA les plus récentes.
Mes domaines d'intervention:
- IA Générative & NLP: Fine-tuning de LLM (classification, chatbot, etc..), prompt Engineering avancé, intégration de pipeline via API, remplissage automatique de formulaires, etc...
- Traitement de données tabulaires: Pipelines d'extraction/normalisation de données, traitement de données hiérarchique et graphes, suppression d'outliers statistiques, ...
- Computer Vision & Deep Learning: Segmentation d'images/vidéos, extraction de caractéristiques, ...
- Déploiement & MLOps: Je conçois des solutions prêtes pour la production. J'industrialise les modèles via Kubernetes, Docker, FastAPI et GCP/Amazon Cloud.
- Activités de recherche: Bayesian modelling/deep learning, variational inference, weight pruning, ...
Mon parcours :
- Actuellement Senior Data Scientist chez Climateseed, je résous une grande variété de problèmes lié au NLP et aux données tabulaires.
- Précédemment R&D Data Scientist chez Tinyclues, j'ai travaillé sur l'amélioration des modèles prédictifs (systèmes de recommandation) et de la stack data.
Französisch
Muttersprachlich oder zweisprachig
Englisch
Verhandlungssicher
Spanisch
Konversationssicher
Nur remote
Führt Projekte hauptsächlich remote aus
Projekt- und Berufserfahrung
- ClimateseedSenior Data ScientistSOFTWARE-HERSTELLERSeptember 2022 - Heute (3 Jahre und 9 Monate)Nice, FranceWorked on a wide variety of NLP and tabular problems:𝗘𝘅𝘁𝗿𝗲𝗺𝗲 𝗠𝘂𝗹𝘁𝗶-𝗹𝗮𝗯𝗲𝗹 𝗖𝗹𝗮𝘀𝘀𝗶𝗳𝗶𝗰𝗮𝘁𝗶𝗼𝗻: I deployed a highly optimized hierarchical classifier to accurately categorize massive volumes of client purchases & compute associated carbon emissions𝗠𝗮𝘀𝘀𝗶𝘃𝗲 𝗗𝗮𝘁𝗮 𝗡𝗼𝗿𝗺𝗮𝗹𝗶𝘇𝗮𝘁𝗶𝗼𝗻: I developed a pipeline combining LLMs and classical data engineering to efficiently unify the extraction of emission factors from highly heterogeneous global databases𝗔𝘂𝘁𝗼𝗺𝗮𝘁𝗲𝗱 𝗖𝗗𝗣 𝗦𝗰𝗼𝗿𝗶𝗻𝗴: I designed a hybrid architecture (semantic AI combined with a deterministic calculation engine) to compute CDP scoresAlso worked on local 𝗟𝗟𝗠 𝗳𝗶𝗻𝗲-𝘁𝘂𝗻𝗶𝗻𝗴 for classification and chatting
- TinycluesR&D Data ScientistSOFTWARE-HERSTELLEROktober 2019 - April 2022 (2 Jahre und 6 Monate)Paris, FranceWorked on improving the predictive model (recommender system), as well as the data stack.𝗧𝗲𝗰𝗵𝗻𝗼𝗹𝗼𝗴𝗶𝗲𝘀 𝘂𝘀𝗲𝗱:- Python (Tensorflow, Keras, Pandas, Parquet, Numba, ...)- Google Cloud Platform (Advanced SQL, Vertex AI, GCS, BQ, ...)- DBT- Kubernetes- Apache Airflow- Terraform- Docker- Amazon Cloud
- Institut MontsourisData ScientistMEDIZINMai 2019 - Oktober 2019 (5 Monate)Paris, FranceI worked on the segmentation and classification of heart scanner images using deep learning. The idea was to evaluate the degree of stenosis of arteries, and generate a health check for each patient.
Empfehlungen
Diese Freelancer passen auch zu Ihren Kriterien
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
Ausbildung und Abschlüsse
- Master ThesisNational University of Singapore2019Courses: ▪ Advanced Deep Learning: MLPs, CNNs, RNNs, GNNs, Attention models, Active learning... ▪ Advanced NLP: Question answering, Sentiment analysis, seq-2-seq, Text classification, POS-Tagging, Dependency parsing... ▪ Computer Vision: Image classification, Segmentation, Image processing, Video tracking, 3D Mapping, Image stitching,... ▪ Uncertainty Modelling: Gaussian processes, Directed and Undirected graph models, Mixture models, Sampling, Variational inference... ▪ Randomized Algorithms Analysis
- Engineer's DegreeEcole Polytechnique2017Engineer's Degree Courses: ▪ Applied Mathematics: Stochastic Processes, Dynamic Models, Optimization under Constraints, ... ▪ Computer Science: Machine Learning Theory, NLP, Computer Vision, Big Data & Database Management, ... ▪ Pure Mathematics: Differential Equations, Real and Complex Analysis, Galois Theory... Long Term Projects: ▪ Deep content based music genre classification (Signal Processing, Deep Learning) ▪ "Can a robot conceptualize a music note ?" (Non supervised learning)