A specialization of Strategy Engines
Industrial Strategy Engines
Where the contest is silicon, energy grids, fabs, fleets, and supply chains, not slide decks. Computational tools for hard physical-system decisions, where the adversary is physics, the regulator, and the clock.
What "industrial" means here
Five domains, one methodological spine: the same Strategy Engines discipline (operations research, simulation, RL, foresight) applied to systems with physical constraints, real-time pressure, and audit trails.
-
Process control under uncertainty
Fab, refinery, grid, batch chemistry. Decisions made under stochastic disturbance, partial observability, and tight safety envelopes. Where classical MPC and ML must coexist.
-
Scheduling at fleet and factory scale
Exact and learned methods combined: GNN-assisted exact solvers, multi-agent coordination for distribution networks, classical OR for hard-deadline cases. Used wherever throughput depends on order, not just speed.
-
Energy and reliability
Reinforcement learning for optimal power flow and stability margins; HPC-backed planning under contingency. For TSOs, DSOs, large-load owners (data-centres, fabs, electrolysers).
-
Compute substrate
EuroHPC, EPI (European Processor Initiative), accelerator programming, intelligent compiler optimisation. Knowing where the silicon stops mattering and the algorithm starts.
-
Supply-chain coordination under geopolitical risk
Multi-agent RL for distribution and supply networks where adversary, partner, and regulator all change behaviour with the news cycle. Decision systems that survive the next sanctions package.
Track record (peer-reviewed)
These are not slide-deck claims. Each anchor below is a published paper with a DOI:
-
High Performance Computing Reinforcement Learning Framework for Power System Control
-
Exact solving scheduling problems accelerated by graph neural networks
Applied case
EquiFlow — NPOO / NextGenerationEU Proof of Innovative Concept (FER): large-scale urban-traffic-coordination optimisation on specialised accelerator hardware. Principal Investigator. Demonstrates the full stack: the OR formulation, the learned decomposition, the accelerator runtime. See the about page for adjacent project leads (VALOR 2025–2029, AnomalyStudio).
Who this is for
- CTO, CTIO, or chief-architect of a deep-tech / industrial-AI organisation evaluating where ML augments and where it must not replace verified control.
- Head of MES, OT, or process-control wanting decision-grade AI that survives audit, regulator, and incident review.
- Strategic-foresight leader in a regulated industry (semiconductor, energy, mobility, defence-adjacent) facing decisions whose blast radius makes ad-hoc judgement insufficient.
- Programme officer or technical board member at an EU-scale industrial-AI initiative who needs an academic-practitioner partner on the methodology side.
Methodology
The core is shared: see Strategy Engines for the framing and the thesis for the longer argument. Industrial Strategy Engines apply the same discipline to problems where the adversary is physics, the regulator, and the time constant of the plant.
Turn Uncertainty into Defensible Advantage
I partner with leaders and institutions ready to solve their most complex strategic challenges.
If your organization is facing a high-stakes problem in one of my three core practices—Strategic Wargaming, Operations Resilience, or AI Governance—let's connect.
Schedule a Strategic Briefing