Automated Cloud Power for Heavy Workloads

KEYSYS built a cloud system for Leeco Steel that runs heavy simulations on demand, uses low-cost cloud resources, and redeploys with a push of a button.

Client Snapshot:

Leeco® Steel is a leading supplier of carbon, HSLA, and alloy steels with 11 locations across North America and Mexico, and is part of O’Neal Industries, the largest family-owned metals service center network in the U.S.

Business Challenge:

The company needed to run large-scale, compute-intensive models every weekday. Their old setup relied on the wrong type of cloud servers, which were powerful but very expensive, and the system was difficult to reset when problems occurred.

KEYSYS Solution:

KEYSYS engineered an Azure-based platform using Terraform, Kubernetes, and GitHub Actions. The system spins up temporary, cost-efficient “spot” servers only when jobs are scheduled, distributes the workload across multiple machines with Ray.IO, and then shuts everything down automatically. Because it’s all defined as code, the entire environment can be rebuilt or updated in minutes.

Impact:

The platform now saves thousands of dollars per month by only paying for compute when it’s needed. Analysts spend time interpreting results instead of troubleshooting infrastructure, and Leeco Steel has a future-proof system that scales with demand.