Freelancer profile translated to English.
Back to original languageAbout Rian
PhD student in NLP at Inria Paris (ALMAnaCH team). I work on LLM pretraining and adaptation.
I published CamemBERT-bio, a French biomedical NLP model with 100k+ downloads on Hugging Face.
During my PhD, I contributed to GAPeron, a suite of open-source French LLMs (1.5B to 24B parameters) trained on +128 AMD/NVIDIA GPUs.
What I can do:
- Pretraining LLMs from scratch (DeepSpeed, FSDP, distributed infrastructure)
- Large-scale data curation
- Deploying 7B-70B LLMs in production (vLLM, GPTQ/AWQ quantization)
- NER and information extraction
- Fine-tuning for specialized domains (biomedical, legal, etc..)
Graduated engineer from ECE Paris and CentraleSupélec in Artificial Intelligence.
Available evenings and weekends for short missions.
Publications (2 first-author, 53 citations):
- CamemBERT-bio (LREC-COLING 2024)
- GAPeron LLMs (submitted to ACL 2026)
- Biomed-Enriched (submitted to ACL 2026)
- CamemBERT 2.0 (arXiv 2024)
Google Scholar: scholar.google.com/citations?user=f5hnXrcAAAAJGoogle Scholar: scholar.google.com/citations?user=f5hnXrcAAAAJ
French
Native or bilingual
English
Fluent
Remote only
Primarily works remotely
Experience
- ViaDialogR&D Consultant - NLP/NERMarch 2025 - June 2025 (3 months)Built French text anonymization system using NER for customer data. Designed LLM annotation pipeline (vLLM + constrained decoding) to generate synthetic training data; distilled to production NER model (0.82 F1).
- PraxysantéR&D Consultant - LLM InfrastructureJune 2023 - June 2024 (1 year)Deployed open-source LLMs (7B-70B) for healthcare applications with quantization (GPTQ, AWQ) on high-end GPUs. Built production inference pipeline with vLLM and FastAPI.
- InriaPhD studentRESEARCHJune 2022 - Today (4 years and 2 months)Paris, FrancePhD thesis on LLM pretraining for clinical NLP, supervised by Laurent Romary and Eric de La Clergerie (ALMAnaCH team).Key achievements:- Trained a 7B decoder from scratch on 128 GPUs (8.7k GPU-hours), matching clinical SOTA with 2.5x fewer tokens- Published CamemBERT-bio, a biomedical NLP model with 100k+ downloads on Hugging Face- Core contributor to GAPeron, a suite of open French LLMs (1.5B-24B parameters)- Research published at LREC-COLING 2024 and submitted to ACL 2026
Reviews
Recommendations
FU
FU
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
- Master's Degree in EngineeringECE Paris École d'ingénieur2022• Maths (Linear Algebra, Probability, Calculus ...) • C/C++, Java, and C# programming • Algorithms and Data Structures, Graph Theory • Advanced Database and SQL
- M2 - Artificial IntelligenceCentraleSupélec2022•Machine Learning fundamentals and Deep Learning •Reinforcement Learning •Computer Vision and NLP •Explainable AI
Certifications
- Developing Android Apps - MOOCGoogle - Udacity2017
- Kotlin for Android Developers - MOOCGoogle - Udacity2018