How ArcelorMittal upskilled a key engineer on AWS IoT Core in five days

Born in 2006 from the merger of Mittal Steel and Arcelor, ArcelorMittal is today the world's leading steel producer, with operations in more than sixty countries on every continent. Its steel goes into automotive, construction, energy, packaging and infrastructure — wherever strength and lightness matter.

The group has made low-carbon transition and industrial digitalization two of its top strategic priorities. This means modernizing its legacy mills: sensors on blast furnaces, predictive maintenance models, supply-chain traceability through unified data platforms.

In 2024 the group committed to a 35 % CO₂ reduction by 2030. That commitment depends in part on every site's ability to stream energy use, steel quality and predictive-maintenance data in real time — hence the urgent need to upskill on modern cloud IoT building blocks.

ArcelorMittal sites have been running on legacy SCADA systems and proprietary historians for decades. Robust, but siloed: each production line generates its own data, in its own format, accessible only from the plant. To drive a group-wide emissions program, that silo model no longer holds.

The group's Digital Industry team made the call: move to a cloud IoT platform able to ingest, normalize and expose telemetry. AWS IoT Core was chosen for its mature MQTT support, its native integration with AWS data services, and its ability to pair with AWS IoT Greengrass at the edge.

To kick off that shift, the group needed an internal technical reference able to frame architectures, challenge integrators and bring the site teams along. That key profile is the one ArcelorMittal handed to Learni — with a clear bar: ship-ready by the end of the week.

Learni designed a one-to-one five-day track, delivered on-site at Learni's Lyon hub. The 1:1 format goes faster and deeper: every concept is anchored on a real ArcelorMittal case, every lab pivots on the learner's questions, without the cadence drag of a twenty-people cohort.

The trainer — an AWS Solutions Architect Professional certified engineer with an industrial-automation background — calibrated the syllabus to real group use cases: blast-furnace data, vibration sensors, energy telemetry. No recycled slides: every lab started from a real plant schematic.

Five days, two flagship labs: (1) connect a simulated device to AWS IoT Core, route it through the Rules Engine into Timestream, surface a QuickSight dashboard. (2) Deploy Greengrass on an edge gateway, run a local inference container, handle the degraded mode under connectivity loss.

By the end of the week, the learner walked out with an operational AWS IoT Core environment, two labs delivered (ingestion + Greengrass edge) and a concrete architecture plan for the first industrial use case. Live satisfaction came in at 100 %.

Beyond the numbers, the value of the one-to-one format shows in the depth of questions asked: it doesn't just cover a syllabus, it also resolves architecture trade-offs that would have stalled the team for weeks if left to figure out alone.