Background
A hardware processing shop running high-mix, low-volume jobs. Frequent changeovers, long setup times, and a workforce that took 6+ months to become productive.
The Pain
- Setup time dominated shift utilization.
- Tribal knowledge lived with 3-4 senior machinists.
- New hires required months of supervised practice before contributing.
What We Built
A process-parameter recommendation model that suggests optimal machine settings per SKU. Combined with a labor-time prediction model for planning. Both deployed on-prem at the factory.
Outcomes
- **Setup time** reduced by 18% on average.
- **Training cycle** for new planners shortened by 67%.
- **Knowledge transfer** no longer depends on the senior machinists' availability.