Automotive manufacturer
Blueprint Computer Vision Pipeline
Forward Deployed Engineer — Palantir
~20,000 blueprints processed for part interchangeability
- Computer Vision
- OCR
- Python
- Palantir Foundry
Problem
An automotive manufacturer had roughly 20,000 parts blueprints and no fast way to identify which parts were interchangeable across them — a supply chain bottleneck when a part shortage meant asking "what else could we use instead?" and not having a quick answer.
Approach
Built a computer vision pipeline to process the blueprint archive at scale, extracting the geometric and dimensional information needed to identify candidate interchangeable parts across the full set of drawings.
Outcome
Processed roughly 20,000 blueprints to surface interchangeable parts, cutting a real supply bottleneck instead of leaving it to manual engineering lookup.