Freelancer profile translated to English.

Description

I design AI systems that understand your documents and automate your processes, deployable on-premise, for companies that do not wish to send their data to cloud platforms.

Specializing in RAG (Retrieval-Augmented Generation) and multi-agent architectures, I help companies harness the power of LLMs in a secure environment.

My approach:

- 100% on-premise AI (Ollama, vLLM) – your data stays with you,

- GDPR & AI Act compliance built-in from the start,

- Granular access control (RBAC/ACL) on documents and AI responses,

- Full auditability (logs, traceability, explainability).

What I do concretely:

- Hybrid RAG systems (semantic + lexical search),

- Orchestrated AI agents (LangGraph, LangChain),

- Secure document ingestion pipelines,

- On-premise LLM integration (Ollama, vLLM).

I speak tech AND compliance (GDPR, AI Act, sensitive data), from POC to deployment.

Clear communication with tech, legal, and business teams.

Last achievement:

Multi-agent EdTech platform (9 specialized agents, RAG + Redis, Reinforcement Learning) - Full architecture from A to Z.

Portfolio: 2501project.dev

Industry field of expertise

Languages

French
Native or bilingual
English
Fluent
Spanish
Conversational

Workplace preferences

Can work on-site

Lyon (up to 50km)

Space Coders Agency
AI Project Manager
March 2025 - November 2025 (8 months)
Lyon, France
• • Design and development of a multi-agent EdTech platform.
• • LangGraph architecture orchestrating 9 specialized agents. These are 9 professors in fields such as Data Analyst, Scientist, Cybersecurity, DevOps, ...
• • RAG pipeline with BM25 Embeddings and Redis cache for optimization.
• • Reinforcement Learning (LinUCB) implementation
• • CI/CD GitHub Actions and real-time monitoring
LangGraph Reinforcement Learning Risk Management RAG Monitoring
Wild Code School
Data Engineering
EDUCATION AND E-LEARNING
November 2024 - May 2025 (6 months)
Lyon, France
• • Design of unified pipelines integrating 4 heterogeneous sources and feeding a platform centralizing Wild Code School's educational data.
• • Prefect and dbt orchestration with hybrid MongoDB / DuckDB storage for OLAP analysis.
• • Development of analytical dashboards (progress, engagement)
• • Data quality automation: anomaly detection, schema validation.
DBT Prefect ETL/ELT Python DuckDB
Wild Code School
Teacher Assistant - Data Training
EDUCATION AND E-LEARNING
September 2024 - December 2025 (1 year and 3 months)
Lyon, France
• • Pedagogical support (courses, livecoding, help, project monitoring, ...).
• • Creation of educational content (courses, exercises, mock exams).
• • Training and skill development (Python SQL, Machine Learning).
Python Machine learning SQL Streamlit Bash