Career
Journey.
From nuclear-fusion simulation and a NASA research lab, through enterprise data platforms embedded inside a regulated client, to production GenAI systems I design, ship and operate myself.
-
Full-time on my own engineering projects rather than client mandates: open source, and production systems I build and operate myself.
- End-to-end, production-grade GenAI systems: hybrid RAG pipelines, AI agents, long-term memory systems, MCP servers, LLM orchestration.
- Production REST APIs and distributed backend services, designed as public contracts rather than internal endpoints.
- Design and operation of a self-hosted distributed infrastructure of 30+ production services — GenAI backends included — with the observability, automation and incident recovery that implies.
- Sole operator of everything I run. There is no one else on call, which is a strong forcing function to do things right the first time.
PythonFastAPIREST APIsDockerLinuxLLMsRAGAI agentsMCPDistributed systemsObservability
-
Three and a half years embedded in the client's teams — four days out of five on site, in a regulated environment — owning architecture and delivery in front of business, compliance and infrastructure stakeholders.
- Tech lead and architecture owner for enterprise-scale data platforms: large Databricks platforms on AWS, ingestion pipelines, Spark workloads, streaming systems.
- Led the Unity Catalog and data governance programmes across two distinct engagements with the same client — different teams, different problem.
- Managed three engineers, then two — a data engineer and two data-visualisation profiles — for roughly eight months to a year between late 2022 and late 2023: backlog, priorities, technical direction, day-to-day guidance.
- Enterprise AI solutions: RAG systems, chatbots, agentic workflows. Governance and supervision of MLflow projects on Databricks with Unity Catalog.
- CI/CD pipelines, automation, monitoring and production reliability in a regulated environment.
2021 → 2023
Engie Smart O&M
Owner of the data catalogue. Migrated the Hive Metastore to Unity Catalog alongside day-to-day operations: around 50 tables per environment, across three duplicated Databricks environments, in about six months. Some twenty people depended directly on that data — IoT telemetry pushed from client sites worldwide — so part of the work was getting the clients themselves to fill in what made the data correct.
2024 → 2025
Engie France Retail
Migrated existing tables to Unity Catalog as external tables, without duplicating the data: around 60 tables in production, over about a year. Fifteen teams of four to five people, on real business domains — billing, electricity contracts, consumption contracts. Change management ran team by team: find the technically motivated person inside each team and work through them rather than mandating from above, then do the final cutover live, with the client watching nothing break.
PythonDatabricksSpark / PySparkSQLAWSUnity CatalogDelta LakeLakehouse architectureTerraformCI/CDDocker
-
- Backend applications and automation tooling; ML tooling and backend APIs.
- Linux infrastructure and self-hosted services.
- Technical consulting and teaching.
PythonLinuxDockerSQLREST APIs
-
- Machine learning models on IoT sensor data.
- Real-time anomaly detection algorithms and time-series prediction.
- Python data-processing pipelines.
PythonMachine learningTime seriesStatistics
-
- C++ software for human behavioural data collection, supporting machine-learning research.
- International research environment.
Earlier research and engineering experiences in Edinburgh and Bristol.
Education
Physics, mathematics, then AI
-
2018 – 2019
MSc Artificial Intelligence
Sorbonne Université (UPMC), Paris
Machine learning, deep learning, statistics, supervised and unsupervised learning, reinforcement learning.
-
2011 – 2016
Engineering degree (MSc equivalent)
Arts et Métiers ParisTech
-
2013 – 2015
MSc Energetic Technology
Universidad Politécnica de Madrid
Thesis in simulation for inertial confinement fusion, at the Instituto de Fusión Nuclear — the work behind the publication.
-
2008 – 2011
Double Bachelor's degree — Fundamental Physics and Theoretical Mathematics
Université de Cergy-Pontoise
Two degrees read in parallel. The mathematics and physics foundation under everything since.
The MSc thesis in Madrid became a published contribution to
Nuclear Theory — the publication.
Certifications
- Databricks Certified Data Engineer Professional (2024)
- Databricks Certified Data Engineer Associate (2022)
- AWS Certified Cloud Practitioner
Languages
- French — native
- English — fluent
- Spanish — fluent
French citizen, EU national. No sponsorship required to work in
Switzerland or the EU.