TA2 - Session industrielle / Industrial Session 1
May 12 2026 10:30 – 12:10
Location: EY (blue)
Chaired by Carlos Zetina
4 Presentations
Progress in GPU-accelerated and Gen-AI-enhanced optimization
In this presentation, we'll present an overview of the latest uses of GPU-acceleration and Gen-AI enhancements to Optimization applications. We'll present some of the strengths and weaknesses of these approaches and what companies need to be aware of before onboarding this latest technology into their stack.
Industrial flexible job-shop scheduling: a matheuristic deployed in production
Scheduling 1000+ tasks on 30 machines in 10 minutes: a print shop's daily challenge. The problem mixes flexible job-shop, precedences, maintenance, machine- and sequence-dependent setups, soft non-overlap, and conflicting objectives. Scale pushed us beyond a pure solver; we present a hybrid multi-round Hexaly + heuristics strategy deployed in production, chasing business acceptability over optimality.
Load Building Optimization via Mixed-Integer Programming: An Industrial Case Study
We present a MIP-based approach to the Load Building problem for an apparel manufacturer, optimizing container filling across distribution centers by maximizing revenue and margin while minimizing stockouts at the customer level. The algorithm iteratively generates batches of homogeneous-quality containers and aims at consolidating similar SKUs to facilitate warehouse operations. Deployed in production for over six months, the approach yields encouraging results.
Smart EV Fleet Charging Solution for the Demand Response Program
Cléo provides smart-charging solutions for operators of electric vehicle fleets. Hydro-Québec offers a program where commercial clients reduce power usage during periods of high demand in exchange for financial credits. We developed a methodology which maximises the credits received from this program while improving the quality of charging solutions.
