Freelancer profile translated to English.

Description

MLOps & AI Engineer – Expert in industrializing and deploying AI at scale. I design and maintain comprehensive MLOps/LLMOps platforms on GCP, ensuring reliable, governed, and monitored deployment of classic ML models, RAG systems, and AI agents. My expertise covers the entire value chain: from data architecture (medallion) and feature engineering, to CI/CD, production observability, and business integration. A technical leader with cross-industry experience (manufacturing, retail, finance), I manage teams and transform AI prototypes into sustainable, high-impact business assets.

Languages

French
Native or bilingual
English
Fluent

Workplace preferences

Can work on-site

Paris (up to 50km)

CASTLEBEE
CTO & Principal Data / GenAI Engineer
DIGITAL AND IT
January 2022 - October 2025 (3 years and 9 months)
Versailles, France
Leading technical strategy (data, MLOps, GenAI), defining engineering standards (code, CI/CD, security), and mentoring consultants on client & R&D projects.
• Leading the technical roadmap and monitoring deliveries for key account client projects.
• Designing data & MLOps foundations: PySpark pipelines, serverless training (Cloud Build), Airflow orchestration, API deployments on Cloud Run / Cloud Functions.
• Mentoring Castlebee consultants, reviewing architectures, implementing best practices for code, CI/CD, and security.
GenAI Projects (VINCI, Société Générale)
• VINCI – SmartBOT: Implementing a RAG document assistant (PDF / DOCX parsing, ada‑002 embeddings), developing a web interface (user/admin spaces), and APIs for integration with business applications.
• Société Générale – Conversational assistant: Scoping and awareness building, implementing an LLM POC (Mistral 7B) with RAG on PDF / DOCX corpus, APIs, and Streamlit interface, managing corpus updates, metadata, and product features (conversation renaming, prompt suggestions).
Global Environment: Python, FastAPI, LangChain, ChromaDB, PySpark, BigQuery, Docker, MLflow, Airflow, GCP, Flask, OpenAI, TensorFlow, Keras, pytesseract, scikit-learn.
MLOps MLflow BigQuery Docker Gitlab CI/CD
Antalis
Data & AI Engineer (GCP)
E-COMMERCE
January 2025 - October 2025 (9 months)
Boulogne-Billancourt, France
MIS BI Department — Modernizing the analytics platform and MLOps practices on GCP (centralizing Oracle & GA4 flows into BigQuery, standardizing data models).
Project 1 — Data Integration Framework
• Designing a packaged Python framework (versioned, documented) for Oracle & GA4 ingestion → BigQuery (ELT) to share data ingestion.
• Managing schema evolution, implementing unit, integration, and data quality tests, observability (logs/alerting), and incident recovery.
• Environment: Python, BigQuery, SQL, dbt, Cloud Functions / Scheduler, Cloud Build, Dataflow.
Project 2 — Analytics Platform & Data Model
• Standardizing the RAW → TRUSTED → ANALYTIC data pipeline, based on dimensional models (star schemas, SCD1 / SCD2) and reusable dbt macros.
• Implementing governance & security (roles, RBAC / IAM, access rules) and Airflow orchestration (DAGs, dependencies, SLAs, alerts, incremental SQL procedures).
• Business visualization via Qlik Sense and technical monitoring via Cloud Monitoring (latency, freshness, completeness).
• CI / CD: Cloud Build templates for dbt, Airflow, and Cloud Run (build, tests, DEV → PROD deployments).
• Environment: BigQuery, dbt, Airflow, Qlik Sense, Cloud Monitoring, Cloud Build.
Project 3 — MLOps: deploying a churn model on GCP
• Preparing data from Oracle via an automated ETL pipeline to BigQuery with full traceability (MLflow).
• Containerizing the model with Docker and deploying it as an API on Cloud Run, orchestrated via Cloud Build (CI / CD).
• Exporting predictions to CRM / reporting, monitoring, and centralizing logs for production performance tracking.
• Environment: Python, BigQuery, dbt, MLflow, Docker, Cloud Run, Cloud Build, Cloud Functions / Scheduler, Dataflow, Qlik Sense.
ArcelorMittal DK
Architect & Data Engineer (Cloudera / Kafka)
RAW MATERIALS INDUSTRY
October 2023 - December 2024 (1 year and 2 months)
Dunkerque, France
Data & Model Team — Digital Transformation Directorate.
Project 1 — Industrial Data Lake (casting line)
• Designing an industrial data lake centralizing PLC / SCADA / LIMS / MES data for real-time analysis and manufacturing process traceability.
• Streaming ingestion via Kafka for sensor/event flows, PySpark processing for consolidation, deduplication, heterogeneous joins, and temporal/geographical enrichments.
• Implementing a medallion architecture (Bronze / Silver / Gold) with robust historization and orchestration on Cloudera / YARN with SLA monitoring.
• Environment: Cloudera, PySpark, Kafka, SQL, GitLab CI / CD.
Project 2 — Vision MLOps Foundation
• Target architecture for the vision MLOps platform (quality inspection, visual recognition) covering image/video ingestion, storage, inference, and monitoring.
• Functional and technical study aligned with the Cloudera / YARN data lake, selecting MLOps components compatible with real-time, resilience, and scalability constraints.
• Production of an architecture document validated by committee, defining a scalable foundation integrating DevOps / CI / CD best practices.
• Environment: Cloud Build, Cloud Run, Docker, Airflow, Kubernetes, GitLab CI / CD, MLflow, PySpark, Kafka, Cloudera.
Apache Kafka MLOps Airflow MLflow GitLab CI / CD