MES replacement across 14 production lines

The problem

The legacy MES had calcified around a single supplier's PLC stack and was blocking the introduction of three new product lines. Engineering changes required vendor service tickets that took 4–6 weeks. Quality data was trapped in flat-file exports — root-cause analysis was impossible at line speed.

How we approached it

• Two-week integration audit covering 230 PLCs, 4 ERP systems, and a homegrown quality database.
• Modular architecture: line-control, OEE, traceability, and quality each own their data and their schema.
• Strangler-fig cutover: new platform ran shadow for 8 weeks before any line was migrated.
• Hot cutover line-by-line over 11 weeks. Each migration included a 24-hour rollback window.

We replaced a system that everyone was scared to touch. Eleven weeks of cutover, zero production minutes lost.

— VP Operations, client-side sponsor

What changed

• Scrap rate dropped 34% within 6 months — driven by real-time SPC alerts the legacy system couldn't surface.
• Engineering changes that took 4–6 weeks now ship in 2–3 days. Plant engineers self-serve through a configuration UI.
• Quality data joined to production data in a single warehouse. First time in company history a recall could be traced to a specific shift in under an hour.

Engagement details

Stack: Go, Kafka, TimescaleDB, Kubernetes, OPC UA, MQTT, Grafana
Duration: 9 months end-to-end (4 months build, 11 weeks cutover)
Team: Solution architect + 7 engineers + 1 SRE + 2 plant-floor liaisons
Engagement model: Full-cycle delivery with 3-year managed-operations agreement

Names and identifying details have been redacted where contractually required. The unredacted case packet — including architecture diagrams, commercial structure, and reference contacts — is available under mutual NDA.

Request the full packet →