Build an Android app that opens the device camera, automatically detects the MRZ or ID card edges, captures a clean image, runs on-device OCR, and extracts all key fields from passports, driving licences and other standard identity cards. Once the text is parsed, the app should immediately post the structured JSON payload (plus the image file) to a local API server I’ll specify at build time. No cloud storage, email or on-device databases—everything funnels straight to that endpoint. CameraX or the native Camera2 API is fine, and I’m comfortable with Google ML Kit, Tesseract, Dynamsoft or any other proven OCR library as long as accuracy is high and the APK stays lightweight. Kotlin is preferred but Java is acceptable if it speeds up delivery. Deliverables • Signed, installable APK and full Android Studio source code • README with build steps, API field map, and any 3rd-party licences • Short demo video (screen capture is fine) that shows a passport and a licence being scanned and the resulting JSON arriving at the local API • One revision round to fine-tune OCR zones or field mapping if needed Acceptance Criteria 1. App launches in <2 s on a mid-range device (Android 10). 2. MRZ lines or ID edges auto-detect; user simply taps once to capture. 3. OCR accuracy ≥95 % on standard-quality documents under normal indoor lighting. 4. JSON schema matches the sample I’ll provide and posts successfully (HTTP 200) within 3 s of capture. 5. No personally identifiable data is stored on the handset once the POST succeeds. That’s the full scope—clean scan, accurate OCR, direct API push.