Project Description
I need a system that lets me drop a set of architectural, engineering, or electrical drawings into a browser (or desktop app) and get an instant, rules-based code compliance report. Code compliance is the first priority, yet the engine must remain flexible enough to bolt on accessibility checks, device counts, spatial analytics, and other smart-building insights later.
Core workflow
• File intake: multi-page PDFs, TIFFs, and CAD exports.
• Recognition layer: OCR and computer-vision routines to read symbols, call-outs, dimensions, and notes.
• Rule engine: evaluate each sheet against local, national, and international building codes I configure.
• Reporting: generate an easy-to-read summary plus a detailed line-item log showing where each rule passed or failed, with sheet and coordinate references.
• Feedback loop: highlight non-compliant zones directly on the drawing preview.
Tech expectations
Ability to run on Ubuntu and Olamma AI
Deliverables
1. Source code with clear project structure.
2. Configurable rule set covering sample local, national, and international codes.
3. Installation script or Dockerfile so I can run everything locally without cloud dependencies.
4. README walking through setup, model retraining, and how to add new code rules.
Acceptance criteria
• Given three sample blueprints (architectural, engineering, electrical), the system flags at least 80 % of the intentional code violations embedded in the test set.
• Reports export to both PDF and CSV.
By delivering these pieces you’ll help me cut plan review time drastically and catch costly design issues long before construction begins.