Freelancer profile translated to English.

Description

Graduated Data Scientist & Data Engineer, with 6 years of experience, I support companies in the design and deployment of Data and AI solutions, from understanding needs to production.

Areas of intervention:

AI & Data Consulting: needs analysis, architecture choices, and feasibility assessment
Industrialization & MLOps: Airflow, CI/CD (GitHub Actions, Jenkins), Docker, cloud
Machine Learning/Deep Learning: classification, anomaly detection, NLP
Data Engineering: ETL/ELT pipelines, SQL/NoSQL databases, scalability (Spark, Elasticsearch, MongoDB)

Experience & projects:

SNCF Réseau (2020–2024): Data Scientist on a digital twin project for the railway network (LiDAR data processing, modeling, industrialization, and deployment of data/ML pipelines)
Personal Project 1: selected by Station F Landing Zone (2026)
Personal Project 2: author of a research paper on the architecture of large language models (LLMs), submitted to the Journal of Machine Learning (Springer, 2025, currently under peer review)

Articles:

Polyakov P. et al, Bayesian monitoring of emerging infectious diseases. PLoS ONE. 11(4): e0152629, 2016. — Institut Pasteur
Polyakov P. et al., Automated force volume image processing for biological samples, PLoS One., 6(4):e18887, 2011. — CNRS

PhD in physics, specializing in chemical physics.

Industry field of expertise

Languages

French
Native or bilingual
English
Native or bilingual

Workplace preferences

Can work on-site

Paris (up to 10km)

Self
The architecture of large language models.
RESEARCH
October 2025 - Today (10 months)
Paris, France
Currently under peer review. Research paper (sole author) on the architecture of large language models. Submitted to Journal of Machine Learning (Springer) on 12/11/2025.
Fine-tuning Glue Hugging Face Transformers LLM
Self
LLM Agent–Driven ncRNA Design via Intrinsic Features and Structure-Guided Feedback
MEDICAL
March 2025 - September 2025 (6 months)
Paris, France
Accepted by Station F Landing Zone on 06/11/2026. Development of two LLM agent-based pipelines for the generation of non-coding RNA sequences, guided by human queries. Each pipeline integrates RNA structural features (BioPython), a 3D structure generative model, and real-time access to scientific literature via PubMed, to foster informed decision-making. This prototype is designed to generate RNA sequences, both de novo and from a template sequence, thus offering flexibility for exploratory approaches.
LLM RAG AI Agent diffusion
Self
Test: 3D Reconstruction of Teeth from Five Intraoral Photographs.
MEDICAL
January 2025 - February 2025 (1 month)
Paris, France
Implementation of the TeethDreamer framework (locally and on AWS for exploring applications in medical practice). Adaptation of the model and re-execution of experiments to evaluate the impact of adjustments on performance.
AWS Cloud Nerf Ubuntu