Rémi Soulignac

Highlights - AI referent across Capgemini entities, consulted on agentic AI strategy and architecture; building and leading an AI tiger team to industrialize the agentification of business processes. - 100x faster incident analysis: designed an AI support agent for a tier-1 WMS, cutting root-cause analysis from 2 hours to 1 minute; MVP rolling out to a 40-person team handling 900 incidents a month. Built an MCP server combining RAG and semantic search, a knowledge graph and an ontology of WMS objects and processes, fed with gold data by a medallion pipeline, plus tools to cross-check live data. - 10x faster delivery, without technical debt: built an agentic harness where specialized agents (architect, developer, tester…) turn requirements into tested, maintainable applications. Lean poka-yoke guardrails, requirement-to-test traceability, requirements engineering and golden-dataset tests against drift (Claude, Copilot, DDD, hexagonal architecture, Rust, Python). - Process agentification: decompose business processes (sourcing, ITIL incident and change management) into activities, inputs, outputs and decision points; select those worth delegating to agents; design multi-agent workflows running them end to end, with humans validating key decisions. Now leading sourcing agentification for a global consumer goods company. - WMS, TMS (Manhattan Active) and OMS architecture and integration for global companies. Sustainability - Architecture study for a global manufacturer to collect and curate ESG data for CSRD compliance. - With IBM: sustainability reporting and continuous reduction of operations' environmental impact. - Reduced the footprint of a large mobile device fleet. - Contributor to Capgemini research on tariffs and future supply chains. Sharing knowledge - Speaker at IFS Connect 2026 on sustainable, cognitive and profitable supply chains. - Community Teaching Assistant, MIT Sustainable Supply Chain Management course (SCM290X), 2023-2026. - Guest lecturer on sustainable supply chains: ENSAM, CNAM, JUNIA.

Critical, highly secured network infrastructures with high availability requirements. Highlights - Led projects of 10,000 person-days: managed 15 engineers directly, several sub-projects, and subcontractors for software and hardened server hardware. - Model-based configuration generator for software-defined networks: replaced manual configuration with a mathematical model (Node.js). Result: 50x faster computation with 50x less code. - Automated test cloud of 100 machines (KVM, custom Linux): one-click multi-VM network environments, automated end-to-end testing and deployment, an early DevOps practice. - Quality by automation for CMMI 3 and ISO 9001: end-to-end traceability between requirements, tests, results, code, bugs and change requests, plus automated delivery notes and audit checklists. Environment: Debian Linux, KVM, Docker, Jenkins, Git, Node.js, Groovy, PowerShell, Robot Framework, JBoss, Apache, PKI, IBM DOORS, Jira, SysML

Supply chain and production information systems for the assembly line: complex planning of component changes in serial production, bill of materials referential. Highlights - Optimize before you upgrade: built a JCL parser, a batch dependency graph and a consumption simulator to optimize mainframe processing for 1 billion parts ordered daily worldwide. Result: a costly mainframe upgrade postponed by 2 years. - Simulation engine to plan small-part changes in serial production, with on-the-fly code generation. - Project management, software architecture and data analysis on strategic supply chain systems. Environment: Mainframe MVS, DB2, JBoss, Spring Framework, Oracle

IT department of an international, multi-site automotive supplier (3,000 employees): infrastructure, software architecture, development and subcontractor management. Highlights - ERP implementation (Axapta, now Microsoft Dynamics): process adaptation in X++ and migration from legacy systems. - 3D mesh generator and viewer for springs and sway bars, feeding finite element analysis (C#, DirectX). - Quote generator with user-defined formulas (C#, dynamic code generation). - Infrastructure built in-house: load balancer for AIX, network monitoring (SNMP), Active Directory roll-out across sites, advanced SQL Server reporting. Environment: C#, C++, Java, SQL Server, Oracle, Reporting Services, IBM AIX, Windows Server, Cisco and HP networking, ASP.NET, Axapta

Advanced C++ coding, following ISO/IEC 14882:1998 standard for a Retail ERP. X86 Assembly coding for critical sections.

IT administration for the French Embassy in Cairo: first international experience.