How We Will Use ML in the SFI Foundation Project

Our SFI Foundation project is about making safety information easier to find and verify.

Right now, users may need to search through static pages to find the right spec or certification record. We want to replace that with a faster, smarter system.

How we will use ML

We plan to use ML in two main ways.

Instead of needing the exact page or exact wording, users can search with a product name, partial spec number, or general term. The ML search can help match that input to the right SFI record faster.

Camera-based equipment recognition

We also want a camera feature that goes beyond QR codes. A user could point their camera at a piece of equipment or take a photo, and the model would try to identify the item and pull up the related SFI data.

Why this matters

  • Inspectors can find information faster
  • Staff spend less time updating static pages
  • Manufacturers work with more current records
  • Users can verify equipment more easily and reduce counterfeit risk

How it fits into the project

ML is one part of the larger system. The project also includes a mobile-first redesign, live certification records, and manufacturer self-service updates.

Our stack is Python/Flask, HTML/CSS/JavaScript, and SQLite. The database stores the information, and ML helps users reach the right record quickly.

In short, we are using ML in a practical way: to make safety data searchable, recognizable, and easier to access.