Built a lightweight, high-impact identity verification service to replace a manual admin process. The system automatically extracts, validates, and processes user ID documents.
The Challenge
New user onboarding was a manual bottleneck. Admins had to physically look at uploaded IDs and type data into the dashboard. This caused a 3-day wait time for users and high drop-off rates.
The Solution
Instead of over-engineering a custom ML model, I initiated a "scrappy" solution using FastAPI and the Google Vision OCR API. I built a wrapper service that accepted ID images, sanitized the OCR output, and auto-filled the admin review form.
⚡ The Impact
slashed user verification time from 3 days to near-instant. Automated 95% of all applications, leaving admins to handle only the edge cases.
System Log End