Full-Stack AI App Engineering

Title

Mobile app (iOS, Android) and web app with AI document processing: client documents to CSV

Project description

Consulting company needs an app where clients submit financial documents. AI reads them, sorts and arranges the data, and produces a CSV file that imports into our software. The app runs on iOS, Android and the web from a single codebase.

How it works

App (iOS, Android, web)

Welcome screen with our logo and a greeting
Start button
Upload screens for each document type: bank statements, ID, payslips and cash statements. Users can take a photo or choose a file (images and PDFs).

Backend

The app sends files to our server over HTTPS.
The server stores the files in Google Drive, with one folder per client.
An AI step reads the pictures and PDFs, extracts the data, and arranges it into the column format our software expects.
The server saves the CSV back to the same Drive folder.

Scope

Cross-platform app (React Native/Expo preferred) with a clean, simple UI
Backend API (Node.js or Python) with login and secure access control
Google Drive integration through a service account
AI extraction using the Claude API or a comparable service. Propose your approach (OCR plus LLM, or a document-AI service) and say why.
CSV generation matching our import format (I will send a sample file or column list)
Deployment to a cloud host with HTTPS, and the web version on Vercel or Netlify
Android installable build and iOS TestFlight build. App Store and Google Play submission support is optional, so please quote it separately.

Security and privacy (mandatory)

These are IDs and bank documents:

User login and authentication on the app and server
Encryption in transit and at rest
Consent screen and privacy policy text
Access controls on Drive folders
Automatic deletion of files after processing, or a configurable retention period
API keys on the server only, never inside the app
Basic OWASP practices, with a short written summary of what you checked
GDPR-aware design, Please state where you would host data and which AI provider processes the documents.

Acceptance criteria

A user can sign in, upload each document type, and receive a processed CSV in Drive on iOS, Android and the web.
The CSV imports into our software without manual fixes.
At least 95% field-level accuracy on an agreed test set of 20 realistic documents, with the correct CSV filled in by hand. Accuracy is the percentage of fields (dates, amounts, names, document type) that match the expected values. Documents the AI is unsure about are flagged for manual review rather than guessed.
Processing time per document set under 60 seconds.
No critical security issues in your final check.

Deliverables

Complete, commented source code in a repository I own
Deployed backend, with environment variables documented
Android build, iOS TestFlight build, and the deployed web version
Setup guide and a simple architecture diagram
A short video of the system working
2 weeks of bug-fix support after delivery

What I need in your proposal

Your proposed AI approach and why
How you will measure and reach 95% accuracy
Where data will be hosted and how you will protect it
Fixed price, milestone breakdown and timeline
2-3 similar projects, ideally document processing or mobile apps with AI
Any questions about the workflow

Timeline: [4-6 weeks]

Suggested skills tags: React Native, Mobile App Development, iOS, Android, Node.js, Python, Artificial Intelligence, OCR, Google Drive API, API Development, Cloud Computing, Web Development