Freelancer profile translated to English.

Description

I help tech teams and innovation departments design and deploy robust AI architectures, from LLM agents to real-time data pipelines, including knowledge graphs and natural language interfaces.

With a PhD from Sorbonne University and 4 years of industrial R&D, I work on complex projects where scientific rigor and production requirements coexist: multi-agent systems, LangGraph workflows, semantic knowledge graphs on Neo4j, and IoT/energy integrations.

What I deliver concretely:

operational AI orchestration platforms, production ML/DL pipelines (PyTorch, HuggingFace), real-time data architectures (Kafka, InfluxDB, PostgreSQL), and knowledge graph reasoning systems.

What sets me apart: I cover the entire cycle from architecture definition to production deployment, with cross-sectoral experience (energy, HR, research) and active contributions to Horizon Europe consortia.

Languages

French
Native or bilingual
English
Fluent

Workplace preferences

Remote only

Primarily works remotely

EU Horizon Marie Skłodowska-Curie Actions
AI Lead MUSIT Project
January 2026 - Today (5 months)
• • Contributing to the MUltiSensorInferred Trajectories project, an international and interdisciplinary EU-funded initiative transforming heterogeneous tracking sensor data (vessels, vehicles, Earth observation) into complete, semantically enriched trajectories.
• • Working on the development of Graph Neural Networks (GNN) for trajectory inference from incomplete and heterogeneous data sources, including images, GPS trajectories, LiDAR and radar data for vessel monitoring.
Deep Learning Machine learning MLOps Python
EU-DREAM, Horizon Europe
AI Lead & Data Platform Architect
November 2024 - Today (1 year and 7 months)
• • Technical lead for the EU-DREAM NLP Intermediator (NLPI) a multi-agent conversational AI system enabling households to interact with their energy infrastructure in natural language, integrated with a Digital Twin layer and deployed across living labs in Ireland and Greece.
• • Designed and implemented the NLPI orchestration layer, coordinating three specialised sub-agents: an Energy Insight Agent, an Energy Optimisation Agent and an Alarm Notification Agent using a hierarchical LangGraph supervisor architecture.
• • Implemented dual-memory management: long-term memory for user preferences and energy profiles, short-term memory for conversation thread continuity, enabling contextually aware multi-turn interactions.
• • Built the underlying data platform (Kafka, InfluxDB, PostgreSQL, MongoDB, Neo4j) and semantic grounding pipelines translating layperson energy queries into structured IoT commands via SAREF ontology alignment and knowledge-graph traversal.
• • Developed LLM-as-Judge evaluation layers, synthetic test-suites and monitoring pipelines for staged rollouts in live energy environments.
Akkodis / GPT-Adecco
Senior AI Engineer
August 2024 - November 2024 (3 months)
• • Design and deployment of an enterprise RAG platform for HR compliance document search, with controlled generation safeguards, evaluation datasets, and audit trails; system in production with internal HR teams.
• • Delivery of a production API detecting discriminatory phrasing in recruitment content; coordination of production deployment with product, legal, and infrastructure teams.