Freelancer profile translated to English.

Description

AI Engineer specialized in Generative AI(LLM, RAG, agents), I support business and technical teams in the design and deployment of Generative AI solutions, from requirement analysis to production.

My approach is based on close collaboration with business teams to thoroughly understand their processes, identify relevant use cases, and design truly useful and adopted solutions.

I am involved in the entire AI project lifecycle:

need analysis, LLM architecture design, development, evaluation, and industrialization.

I have notably worked on:

Business document assistants based on advanced RAG
Systems for querying data in natural language (Text-to-SQL)
Multi-step LLM agent workflows to automate complex tasks
Cost/performance evaluation and optimization pipelines for LLM systems

These solutions have been deployed in production in cloud and on-premise environments, with strong constraints (finance, payments, fraud).

Industry field of expertise

Languages

French
Native or bilingual
English
Fluent

Workplace preferences

Can work on-site

Paris (up to 50km)

BPCE Payment Services
AI Engineer / Generative AI
BANKING AND INSURANCE
April 2023 - Today (3 years and 2 months)
Charenton-le-Pont, France
Design, development, and industrialization of generative AI systems in production on GCP and constrained on-premise environments.

Multi-domain Text-to-SQL (Natural Language Querying)

Design of a generalizable solution allowing non-technical profiles to query their data autonomously

• Multi-RAG architecture: business documentation + BigQuery schemas + few-shot examples
• PostgreSQL vector indexing (pgvector) with semantic search + metadata filtering & rights management
• SQL generation via LLM (GPT, Gemini) + open-source fine-tuning experiments
• BigQuery pre-execution pipeline: cost estimation, error detection, LLM auto-correction
• Continuous evaluation: user feedback, Schema Linking Accuracy
• Deployment on GCP (Cloud Run, GKE, Vertex AI)
• Facilitation of Prompt Engineering workshops for business adoption

Stack: LangChain, Haystack, GCP, BigQuery, PostgreSQL/pgvector, Python

Payment Assistant – Advanced RAG

Development of an assistant centralizing technical, regulatory, and business documentation.

• Multi-step RAG pipeline: query analysis, hybrid retrieval, cross-encoder reranking, parent-child expansion
• Advanced document parsing (Azure Document Intelligence) with hierarchical reconstruction
• Table & image processing via multimodal LLM
• Evaluation pipeline: DeepEval, Ragas, LLM as a Judge + user feedback
• Full industrialization on OpenShift with CI/CD (Jenkins, Artifactory, XL Deploy/Release)

Stack: LangChain, Azure DI, OpenShift, Jenkins, Python, React
Prompt Engineering AI Agent RAG
Microsoft
Final Project for ALIS Association
October 2022 - March 2023 (5 months)
Design of an augmented alternative communication prototype for people with Locked-in Syndrome
Implementation of conversational NLP models for context-aware response generation and text prediction
Experimentation with BERT, GPT-2, and DialoGPT for language adaptation to the discussion context
Integration of a user interface, preceded by UX work, focused on accessibility and personalization