Guillaume Marques

GRDF is the main distributor of natural gas and biomethane in France. GRDF operates the largest gas distribution network in Europe (over 200,000 km of pipelines, 9,500 municipalities covered) and serves 11 million customers.

I joined IA Fabric to develop and industrialize software for optimizing gas distribution network equipment settings and simulation. I collaborate with a Data Scientist on the algorithmic aspects, while ensuring integration into the IS in conjunction with Data Engineers and the Project team (PO/CP).

My contributions:

- Technical scoping and backward specification of business needs

- Design and development of the optimization and simulation software, deployable nationwide

- Delivery of production-ready code

- Development of the validation strategy to ensure reliable decisions

- Comparative experimentation campaigns against the legacy software (source of truth)

- Collaboration with a researcher specializing in numerical optimization of energy infrastructures.

- Writing scientific and technical documentation: specifications, mathematical model, algorithms, comparative analysis

Technical context: Julia, JuMP, Ipopt, MadNLP, HiGHS, SCIP, Gurobi, GAMS, Python, C++, CI/CD, unit tests, e2e tests

Atoptima is a deeptech startup, a spin-off from Inria, specializing in optimizing complex supply chain operations planning, offering SaaS solutions for industrial and logistics companies. I worked in a team of two operations research engineers, a back-end engineer, and a front-end engineer.

My achievements:

- Full-stack design and development of products (operations research, front-end, back-end, database, deployment)

- Lead developer of the open-source packages Coluna, DynamicSparseArrays, and BlockDecomposition (totaling 220+ GitHub stars)

- Recruitment and management of the first four engineers

- Contribution to fundraising (€1.2M): product development and commercial showcase

Startup website: Coluna GitHub:

Technical context: Julia, Rust, JuMP, Gurobi, TypeScript, Angular, NestJS, SQL, NoSQL, PostgreSQL, AWS, Kubernetes, CI/CD, unit tests, e2e tests

I conducted research and development on exact mathematical optimization methods for two-level vehicle routing problems, applied to urban logistics. I was part of the Inria Realopt team, internationally recognized for its work in combinatorial optimization and vehicle routing.

My achievements:

- Finalist for the VeRoLog Doctoral Dissertation Prize (European award for the best thesis in routing optimization and logistics)

- Development of the most efficient exact method in the literature for two-level vehicle routing problems, solving 100% of open instances with capacity constraints and 90% with time windows

- Publication of three articles in leading journals (Computers & Operations Research, Transportation Science, Operations Research), totaling 160+ citations, including one article with 90+ citations

- International presentations: VeRoLog, ISMP, JuMP-dev, ROADEF, SPOC school 2020 on "Autumn school on Advanced BCP Tools: VRPSolver and Coluna"

- Initiation of the Coluna package, a Branch-cut-and-price framework in Julia

Technical context: Julia, JuMP, C++, CPLEX, HTML/JS