DSA — Data Science & AI Solutions
Models that read the asset, not the report.
We build machine learning for physical infrastructure — pipelines, plant, and the sensors already attached to them. The work is judged on whether it changes where an inspection crew is sent tomorrow.
Work
PipeGuardian AI
A proposed deep learning system that reads pipeline sensor data in real time, locates erosion, and classifies what kind it is — so inspection crews are sent to a place rather than along a route.
Read the case →- Inputs
- Ultrasound, LiDAR, pressure sensors
- Analysis
- Real-time and historical, cross-validated
- Output
- Location, erosion class, failure forecast
- Target accuracy
- 95%+ vs 70–80% conventional
How we work
Most industrial data problems are not modelling problems. They are integration problems wearing a modelling costume: three sensor systems that do not agree, a historian nobody has queried in years, and a maintenance log kept in a spreadsheet. We start there.
SKELETON — this division currently has one case study and no service pages. Whether PEC sells DSA as consulting engagements, as products, or as both is still an open question. See PLAN.md section 7.