truuimage
Forensic AI-image detection, shipped as a complete SaaS.
- Role
- Sole Developer · Full-Stack SaaS
- Year
- 2026
- Stack
- Next.jsVercelStripeREST APIHashed API keysTOTP 2FAAsync scan pipelineDirect-to-storage uploads
A calibrated verdict on whether a photograph is real, AI-generated or manipulated, with the evidence behind it, built for journalists, moderators and investigators who need a defensible read before they act on a photo. I built the product end to end: the four-stage detection pipeline, the credit billing, the public REST API and the web console around them.
- 01
Built the entire product end to end, detection pipeline, billing, API and web console.
- 02
Four-stage scan pipeline: global neural scoring, regional analysis, statistical forensics across compression, noise, texture and frequency, then a calibrated fusion into a three-class verdict with a confidence tier.
- 03
Credit billing on Stripe, credit packs rather than seats, a credit ledger, and an automatic refund when a scan fails or expires.
- 04
Public REST API with hashed API keys, detect and poll endpoints, error codes and worked examples in curl, JavaScript and Python.
- 05
Accounts with email verification, TOTP two-factor and a self-serve deletion flow with documented data retention.
- 06
Uploads go straight from the browser to storage; originals are deleted within seconds of a scan, keeping only a 512px re-encoded thumbnail with EXIF and GPS stripped.