Freelancer profile translated to English.
Back to original languageAbout Sara
AI Doctor and researcher, I help ML and R&D teams diagnose, validate, and optimize their deep learning models and LLM-based systems.
I intervene when training is unstable, too costly, difficult to reproduce, or when performance degrades after a change in the model, data, or pipeline.
My areas of intervention include:
— Audit of training and fine-tuning pipelines;
— Diagnosis of instabilities, divergences, and optimization problems;
— Reduction of GPU costs, computation time, and memory usage;
— Evaluation and adaptation of methods from scientific publications;
— Design of benchmarks, experimental protocols, and ablation studies;
— Development, debugging, and analysis of models in Python and PyTorch;
— Optimization of training on GPU and HPC infrastructures;
— Analysis of the quality, robustness, and reproducibility of results.
I work at the intersection of deep learning, LLMs, optimization, high-performance computing, and applied research. My background has led me to contribute to projects in scientific AI, video analysis, computational modeling, and large-scale learning systems.
I favor short, focused, and high-technical-value missions, with directly usable deliverables: audit report, documented diagnosis, prioritized recommendations, experimental plan, reproducible prototype, comparative benchmark, or technical presentation.
Key skills: Deep Learning, Machine Learning, LLM, PyTorch, Python, fine-tuning, model optimization, benchmarking, distributed training, GPU, HPC, and scientific experimentation.
French
Native or bilingual
English
Fluent
Remote only
Primarily works remotely
Experience
- RIKENAI and High-Performance Computing ResearcherFebruary 2025 - December 2025 (10 months)Tokyo, JapanLed applied research projects at the intersection of artificial intelligence, high-performance computing, and large-scale scientific computing. Developed machine learning methods and experimental pipelines for various domains, including computational immunology, video analysis, and scientific data exploitation.Worked on optimization problems in stochastic and non-convex contexts, as well as efficient training strategies for modern deep learning models. Analyzed challenges related to architectures and systems for large-scale AI, including routing, communications, scaling, and performance on intensive computing infrastructures.Designed, trained, evaluated, and diagnosed deep learning models on GPU and HPC infrastructures, in collaboration with interdisciplinary research teams. Implemented reproducible experimental protocols, analyzed trade-offs between performance, computational cost, and scalability, and formulated technical recommendations based on the results obtained.
- Eiffage Energie Systèmes Vélizy-VillacoublayProject Manager and Data Scientist -- AI / IoTNovember 2019 - December 2024 (5 years and 1 month)Vélizy-Villacoublay, FranceManaged the R&D project “Future Road,” conducted with several industrial partners on intelligent instrumentation of road infrastructure.Designed a roadside data collection unit combining cameras, lidars, and infrared sensors to measure road wear and various traffic parameters. Coordinated the integration of equipped mats with nano-sensors for real-time vehicle weighing.Supervised a team of software developers at Eiffage and contributed to the data science, data processing, and system integration components. Supported the project until 2023 to ensure its technical continuity and deployment.
Recommendations
Be the first to recommend Sara
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
- PhD in AI and HPCUniversité Paris Saclay2024Ph.D.
- Master 2 in Computer Science -- High-Performance ComputingUniversité Paris Saclay2019