Case 01 · Computer vision · Full-stack

SakhiSign

An AI platform that tells you whether you signed it right — and which part was off.

Role
Architected, built and deployed
Stack
React · Vite · Node.js · Express · MongoDB · FastAPI · MediaPipe Hands · Dynamic Time Warping
Links
Live demo ↗ Code ↗

The problem

Checking whether a sign-language gesture was performed correctly usually means training a classifier on a large labelled dataset — one that has to grow every time a new sign is added. And a plain pass/fail doesn't tell a learner what to fix.

What I built

  • A zero-training-data evaluation engine: each sign needs just one reference recording.
  • Real-time tracking with MediaPipe Hands — 21 landmarks per frame turned into an 18-D feature vector.
  • Dynamic Time Warping to compare a learner's attempt against the reference, even at a different speed.
  • Mirror-invariant scoring and targeted, per-component feedback on each attempt.
Evaluation engine & architecture
Webcamlive frames MediaPipe Hands21 landmarks / frame 18-D vectorper frame DTW alignmentspeed-invariant Feedbackper component Reference1 per sign · mirrored 3-tier architecture React · Vite clientcamera + feedback UI Node · Express APIJWT auth · MongoDB FastAPI serviceinference

Key decisions

21landmarks tracked per frame
18-Dfeature vector per frame
0training samples needed
3tiers, deployed live
Next caseLead Scoring System →