Build a Self-Hosted OCR + Face Recognition + Liveness Web API (No External Dependencies)

Project Description

We are looking for an experienced developer/team who has already built a Document OCR + Face Recognition + Face Liveness solution.

We need a self-hosted web application (API-based system) that will:

Receive document images (Aadhaar, PAN, etc.) from a mobile app via API

Perform OCR and extract structured data

Detect face from document

Capture and verify selfie image

Perform face matching (document photo vs selfie)

Perform face liveness detection

Return processed results in structured JSON format to the mobile app

Architecture Requirement

Mobile App → Sends images via API

Web Application (Hosted on our server) → Processes OCR + Face + Liveness

Returns verification result via API

Important:
The system must be fully self-hosted.

No third-party paid APIs (No AWS, Azure, Google Vision, etc.)

No external runtime dependencies

All processing must run on our own server

We should have full source code ownership

Open-source libraries are acceptable.

Expected Features

High accuracy OCR for Indian ID documents (Aadhaar, PAN)

Structured data extraction (Name, DOB, ID number, etc.)

Face detection from document

Face embedding generation

Face similarity scoring

Passive liveness detection

JSON-based API response

Scalable architecture

Preferred Tech Stack (Flexible)

Backend: Python (FastAPI preferred) / Node.js / .NET

OCR: Tesseract / PaddleOCR / Custom model

Face Recognition: InsightFace / FaceNet / DeepFace

Must support containerized deployment (Docker preferred)

Important

Please apply only if:

You have already built a similar system

You can share demo or past work

You understand document OCR + face AI pipeline

You can start immediately

We are ready to begin this project quickly.