MLOps & AI Engineering

Nelson Rogers

AI Platform Lead, MLOps and agentic systems (Groupe Flow Line). I take AI from PoC to production: architecture, industrialisation, reliability, security.

Education Top of my class ISIS, INSA Group
Certification Google Cloud Professional Data Engineer
Patents Co-inventor Two patents, US and Europe

Former competitive alpine skier.

See the projects Career

ProfileCurrent

Nelson Rogers

Rhône-Alpes or fully remote

role: AI Platform Lead, MLOps and agentic systems
employer: Groupe Flow Line, Lyon
education: Engineer, ISIS (INSA Group), top of graduating class
certification: Google Cloud Professional Data Engineer
patents: US2024412073A1 · EP4475044B1

LinkedIn GitHub Contact

Selection

Recent projects

View the portfolio

01 · Flow Line Since January 2026

Supply chain AI platform · Flow Line

Owner of the platform that plugs into clients' ERP (Sage X3): sales forecasting, scheduling, anomaly detection, replenishment. Industrialisation (Prefect, MLflow) and agentic layer (ADK, MCP).

Prefect
MLflow
ADK
MCP / FastMCP
Sage X3

Read the project

02 · APERAM · Flow Line June 2026

Agentic scheduling · APERAM

Agentic layer (ADK) letting planners drive an OR-Tools optimisation engine in natural language, with full traceability and human validation.

ADK
AI agents
OR-Tools
LLMOps

Read the project

11 · Atos Trophy 2023

MLOps patents · Atos Inventor Trophy 2023

Co-inventor on two patents (US and Europe) on selecting training datasets through reinforcement learning. Winner of the 2023 Atos Inventor Trophy.

US2024412073A1
EP4475044B1
Reinforcement learning

Read the project

Architecture

From the ERP to AI agents

The supply chain AI platform whose architecture I own at Groupe Flow Line: a shared layer that ingests and normalises ERP data, domain services, and every capability drivable by AI agents.

Pipeline · Supply chain AI platformGroupe Flow Line · Lyon

01 ERP Sage X3
→
02 Ingestion Prefect
→
03 Models MLflow
→
04 Services per domain
→
05 Agents ADK, MCP

services: Sales forecasting · Production scheduling and planning · Anomaly detection and customer prospecting · Replenishment optimisation
connector: Sage X3 as the reference, with an ERP-agnostic architecture
design: Multi-tenant, strict isolation of each client's data
guardrails: Human validation before any action with side effects, full traceability

Tools

Tech stack

Full skill set

MLOps01

Prefect
MLflow
Airflow
Kubernetes
Docker
CI/CD
Terraform

AI / ML02

AI agents (ADK, MCP)
LLMs
RAG
Classical ML
Reinforcement learning
OR-Tools

Cloud03

Azure
AWS
Google Cloud (PDE certified)
On-premise

Contact

Let's talk

I'm open to discussing MLOps/AI opportunities, consulting or collaborations.

Contact me Download the portfolio (PDF) References