Philippe Baus Product engineer

CV

After four years as a rail traffic controller at SNCF, I went back to school at 23: a two-year computer science degree, then an engineering degree at ENSEEIHT as a work-study student at Capgemini Engineering, with excellent results at every step. Hired full-time as a Java developer on the 3DEXPERIENCE PLM platform, I quickly contributed to the core framework and to the technical design of features, becoming tech lead of a team of 15 in under two and a half years. Training, Anthropic certifications and AI projects followed: today I am a product engineer at SFEIR, building applications that use AI while developing with it myself.

1. Product Engineer · SFEIR

   08/2026–present

   AI agents working on industrial data foundations.

   • Custom MCP servers and tools for Claude and Gemini: API calls, extraction.
   • Neo4j knowledge graph, plain Cypher queries.
   • Prompt engineering · structured instructions and targeted examples to make model answers reliable.
   • Context engineering · each call only gets the relevant context, drawn from the graph and the tools, to cut noise and hallucinations.
   • Eval suites · hybrid evaluation pipeline, code-based automated tests, deterministic metrics and LLM-as-a-judge.
   • Guardrails · least privilege, fine-grained permissions through API tokens, tightly restricted tool scope.
   • Human in the Loop · mandatory human approval for any critical operation or data change.
   • AI-assisted development · Claude Code kept in check: permission modes, blocked access to secrets, vetted skills and MCP servers, TDD, automatic linting and formatting.
   • Technology watch · already a habit, now a daily one in the AI era: new models, new AI tools and security vulnerabilities, essential to follow for clients as well as for the services I host at home.

2. Tech lead · Capgemini Engineering

   02/2026–07/2026

   Technical foundation of a custom PLM, team of 15 developers.

   • Requirements analysis · a project started with no technical specification: detailed functional specifications with heavy business logic, turned into technical specifications developers can act on.
   • Link with decision-makers · direct reporting on quality and progress.
   • Technical foundation · shared Java Jakarta codebase for 3DEXPERIENCE widgets, selection and validation of libraries, unit testing standards.
   • Methodology · wrote the project’s development methodology.
   • Team · technical leadership and infra and dev support for 15 people, setup and maintenance of development environments, delivery assurance.
   • Integration · widgets connected to the 3DEXPERIENCE platform through REST APIs and the 3DX Java SDK.
   • Complete GitLab CI pipeline · build, unit tests, OWASP Dependency-Check security analysis, automated RPM packaging with install scripts.

3. Java / Spring PLM Developer · Capgemini Engineering

   03/2024–01/2026

   EDF Switch project: I joined as a junior developer and left taking part in decisions on the project’s structure.

   • REST APIs and Vue.js widgets on 3DEXPERIENCE; SOLID, Clean Code, test coverage > 90%, code reviews and mentoring.
   • Client need · direct back-and-forth with the EDF project managers, who owned the expression of the need.
   • Foundation shared by 10 teams · framework for wrapping Web Services and for data-oriented processing, with a modeling diagram I proposed that the teams adopted.
   • By the end of the project · helped the architects decide on the project’s structure, abstractions and how information was passed on to the teams.

4. PLM Training & Data Migration · Capgemini Engineering

   10/2023–03/2024

   Internal training module on 3DEXPERIENCE widget development and PLM data migrations.

5. Apprentice Engineer · Capgemini Engineering

   10/2020–10/2023

   Two separate projects during my work-study program.

   PLM data migration

   03/2023–10/2023

   For a client: started from scratch and taken to production, as lead developer under the responsibility of a project manager.

   • ETL chain from a legacy CAD tool to a more recent one, for more than 1,000 PLM objects, using only the tools already installed at the client: Visual Basic, PowerShell, VBA.
   • Orchestrator and unified logging on all platforms; audit then optimization: −20% processing time.

   Connect

   10/2020–07/2022

   Internal asset: an abstraction for interconnecting PLM systems.

   • Critical module detecting concurrent imports, in Java and Neo4j, to prevent data corruption.

6. Python & Data Developer · INSA Toulouse

   04/2020–07/2020

   Energy optimization of wastewater treatment plants.

   • Cleaning and harmonizing data from heterogeneous sensors.
   • Mathematical models (Pandas, NumPy, SciPy) and a visualization interface.

7. Rail Traffic Controller · SNCF Réseau

   09/2014–08/2018

   Safety and punctuality of rail traffic.

   • Station control panel (TCO) supervision, strict safety procedures.
   • Operational rigor and stress management.