À propos de Paul
Specialized in advanced image & video processing and deep learning models, from scalable training to optimized inference for edge deployment, for a diverse range of applications, from aging prediction algorithms on iOS devices to real-time video processing on edge-device, with a focus on numerical fidelity, high throughput and seamless integration. Experienced in developing and validating AI-driven algorithms with strong skills in computer vision, model optimization, and real-time implementation aligned with industry standards. Keenly interested in AI embedded applications for robotics, defense, aeronautics, and industry 4.0
Français
Bilingue ou natif
Anglais
Capacité professionnelle complète
Roumain
Bilingue ou natif
Accepte de travailler sur site
Paris (jusqu’à 50 km)
Expériences
- Safran.AIEDGE DEEP LEARNING ENGINEER (FREELANCE)DÉFENSE & ARMÉEjuin 2025 - octobre 2025 (4 mois)Paris, France• Deep learning model optimization mission for GPU-based edge device deployment with real-time video processing constraints.• Achieved up to x4 increased model inference throughput and decreased memory consumption, as well as less than %1-5 loss across task performance metrics, with successful integration into a video pipeline stack.• Developed in Python, Torch, ONNX, TensorRT, Triton, FFmpeg and Gstreamer
- L'Oréal SA - L'Oréal FranceDEEP LEARNING ENGINEER (FREELANCE)MODE & COSMÉTIQUESnovembre 2023 - octobre 2024 (11 mois)Paris, France• Engineered a deep learning and computer vision model for the Product Evaluation team, resulting in a 95% match in humanprecision for mascara product assessments, enabling faster decision-making on dozens of new products per quarter.• Contributed to neural radiance field and ’deep fake’ upstream studies for volunteer anonymization in videos, while executingvideo text extraction using LLMs that improved the analysis of customer reviews, aiding in the development of future RAGs• Oversaw the engineering and management of simultaneous model training experiments and validation under deadlines andcost constraints, as well as demonstrations for Q&A, as well as their optimization and minimization.• Developed in Python, Torch, Lightning, Hydra, DeepSpeed, ONNX, Streamlit, GCP (Cloud Storage, Vertex AI), Hugging Face.
- L'Oréal SA - L'Oréal FranceEDGE ML ENGINEER (FREELANCE)MODE & COSMÉTIQUESseptembre 2022 - novembre 2023 (1 an et 1 mois)Paris, France• Retro-engineering of custom Bayesian "black-boxed" models for aging prediction on iOS devices.• Validated the accuracy of the custom implementation by achieving a numerical match of 99.999% with the black-box results, and ensured seamless integration into the edge device environment on time for delivery deadlines.• Developed in Python for prototyping, and C++ with Dlib for iOS integration.
Recommandations
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
- Bachelor of ArtsUniversité Paris Descartes2019Bachelor ofArts in Computer Science,
- Master of Professional StudiesIA School2022Professional Master in Data Science,