Description

I ship autonomous AI agents and RAG pipelines: from prototype to production.

At Brevo: multi-agent AI handling documentation Q&A, CRM operations, and campaign analytics. Built with LangGraph / Google ADK, served via FastAPI, deployed on Kubernetes, monitored with Langfuse. LLM-as-a-judge evaluation, human-in-the-loop workflows, and fine-tuned BERT for multilingual spam detection. Exposed the entire API as MCP servers for AI coding assistants.

Before that: led ML engineering at Forvia (predictive maintenance, manufacturing quality, time series forecasting). At Orange: applied PhD research to anomaly detection on telecom networks at scale.

PhD INSA Lyon: transformer-based unsupervised anomaly detection. Published papers (ECML-PKDD, IEEE), two international patents.

What you get:

- Autonomous AI agents (LangGraph / ADK) with tool calling and memory

- Production RAG pipelines with evaluation and quality gates

- Fine-tuned models (LoRA/QLoRA, BERT, vLLM serving)

- Deployed microservices (FastAPI, Docker, K8s)

- Full observability: Langfuse, tracing, automated checks

Stack: Python · LangGraph · Google ADK · LangChain · Langfuse · FastAPI · Docker · Kubernetes · GCP · vLLM · HuggingFace Transformers · Spark · SQL

Languages

French
Native or bilingual
English
Fluent
Arabic
Native or bilingual
Spanish
Conversational

Workplace preferences

Can work on-site

Paris (up to 50km)

Brevo
Senior AI Engineer
September 2024 - Today (1 year and 9 months)
Paris, France
Built and deployed a multi-agent AI assistant in production (LangChain → LangGraph → Google ADK migration), with agents for documentation Q&A, CRM management, and campaign analytics Exposed Brevo's public APIs as MCP servers (FastMCP), enabling AI coding assistants (Claude, Cursor) to use Brevo tools natively
Implemented production-ready RAG pipelines (RagFlow, Neo4j) to retrieve information from the Brevo Knowledge Base and answer customer queries Integrated Langfuse for end-to-end tracing and automated quality checks via LLM-as-a-judge. Trained and deployed ML audience recommendation models (MLOps with ZenML and Python) to optimize targeting and maximize e commerce campaign ROI. Fine-tuned a pre-trained neural network (BERT) to detect spam and classify fraudulent email patterns. Implemented scalable API endpoints (FastAPI) to serve AI microservices (image generation for email campaigns, speech recognition for phone calls)
LangFuse Google ADK RAG GenAI LangGraph
Forvia
Senior Data Scientist - Technical Leader
February 2023 - September 2024 (1 year and 7 months)
Paris, France
Developed Innovative Data-Driven Solutions in Automotive Manufacturing: Machine Learning Machine predictive maintenance, process parameter optimization, time series forecast, recommendations. Selected as a Machine Learning Expert
GenAI Time Series Analysis MLOps Spark Google cloud
Orange SA - Orange Digital Ventures
Data Scientist
TECH
January 2019 - January 2023 (4 years)
Orange, Grenoble, France
Explored machine learning-based anomaly detection, applied to network monitoring Published papers at international conferences Submitted two patents to INPI
GenAI Pytorch Anomaly Detection Python Kubernetes