Waste reduction in hard-drive manufacturing
We applied machine learning to a hard-drive production line to understand - and reduce - the waste that high-precision manufacturing generates.
Manufacturing MLRoot-cause analysisDeployment
// The challenge
Where it started.
Hard-drive manufacturing is unforgiving: tiny process drifts create scrap at scale. The causes are buried across high-dimensional sensor and process data.
// Our approach
- 01
Consolidated process and sensor data into a usable feature set.
- 02
Built models linking process conditions to defects and scrap.
- 03
Surfaced the drivers engineers could actually act on.
- 04
Deployed monitoring so issues are caught before they propagate.
// The outcome
What it delivered.
root-cause
of scrap, found
less
waste at source
live
line monitoring
The drivers of scrap made visible and addressable.
Process adjustments that cut waste at the source.
A monitoring capability that keeps the line in tolerance.
// More case studies
