From Graph to Edge: Designing Distributed, Reusable Industrial AI

DotVision, which builds embedded AI and IoT systems already running in aerospace, energy and rail, presents the distributed AI architecture it is now industrialising. Two constraints shaped it: labelled fault data is rare or nonexistent, and raw data can't leave the site. The design is a shared executable graph that models, simulates, orchestrates and deploys intelligence down to the microcontroller, combining physical models, signal processing, business rules and AI experts that switch on only when needed. The baseline is learned on a healthy machine, with no fault examples. The demo is a full-chain proof of concept diagnosing a rotating machine. Thursday 3 December, 18:00 to 20:00, at datacraft, 3 Rue Rossini, 75009 Paris. In French. Free, approval-gated, about 30 places. Why go: a concrete edge-AI architecture from a team with sensors already in production. Best for engineers working on industrial IoT, predictive maintenance or embedded ML.