Freelancer profile translated to English.

Description

- Data Scientist NLP & Generative AI | Industrialization & MLOps Expert

- Specialist in deploying large-scale LLM and RAG solutions on Cloud, I manage the optimization of complex AI architectures to reduce inference costs.

- Passionate about technical excellence, I combine algorithmic mastery (Python, Transformers) with MLOps practices to transform unstructured data into growth levers and accelerate decision-making in critical environments.

Languages

Chinese
Native or bilingual
French
Fluent
English
Fluent

Workplace preferences

Can work on-site

Paris (up to 50km)

Groupe Covéa (MAAF, MMA et GMF)
Data Scientist / AI Engineer
BANKING AND INSURANCE
January 2025 - January 2026 (1 year)
Paris, France
1. Managed the design and deployment of ML, LLM, and Agent projects: fraud detection in Home Insurance (multi-risk), customer verbatim analysis, and BI report generation with RAG and LangChain.
2. Industrialized frugal AI solutions in production, monitoring cost/latency by orchestrating Azure, Databricks, and OpenAI services to reduce inference costs by 31% and ensure scalability.
3. Translated business needs into technical solutions and developed tools to aid in risk claim assessment (via a Streamlit + Azure DB application), increasing team analytical efficiency by 30%.
LLM Generative AI Property and Casualty Insurance Deep Learning Workflow Automation
Groupe Covéa (MAAF, MMA et GMF)
Data Scientist (End-of-Studies Internship) - DSIN
BANKING AND INSURANCE
March 2024 - September 2024 (6 months)
Paris, France
Comparative evaluation of solutions for multi-label classification of customer verbatim, including open-source LLMs, GPT, and a CamemBERT fine-tuned on Azure Databricks, automated with LangChain and MLflow.
Development of advanced prompting architectures (parameterization, auto-prompt generation, and majority voting) for GPT-3.5-Turbo, improving the classification F1-score from 0.60 to 0.79 (results published on Medium).
LLM Extract, Transform, Load (ETL) Generative AI Deep Learning Hugging Face
Outlier AI
AI Consultant
July 2024 - Today (1 year and 11 months)
Evaluation and improvement of large-scale language models (LLMs) on complex tasks, including logical reasoning, code generation (Python), and mathematical problem-solving. Writing 'golden responses' and precise instructions to guide model fine-tuning and alignment with human intent. Collaboration with research and engineering teams to provide detailed qualitative feedback on model behavior.
IT Project Management LLM Generative AI RLHF Evaluation