OCR Web App MVP

I need an end-to-end, web-based MVP that lets users upload single or multiple PDF files—both scanned and digital-born—then runs them through Google Cloud Vision to extract text, tables, and any clearly tagged text blocks with metadata. After extraction the app must highlight differences when two versions of a document are compared, flagging added, removed, or altered content in an intuitive “diff” view.

A clean, modern, responsive UI is essential. Users should be able to review the results on screen and export everything—raw OCR text, structured tables, comparison report—to Excel or a neatly formatted PDF. All functionality must work smoothly on common desktop and tablet browsers.

Key expectations
• Secure PDF upload, queuing, and progress feedback
• Accurate OCR via Google Cloud Vision, including table recognition
• Reliable comparison workflow that pinpoints mismatches
• Clear frontend built with a mainstream framework (React, Vue, or similar)
• Well-structured backend (Node, Python, or comparable stack) with tidy code comments
• Containerised or scripted deploy instructions
• Source code handed over at each approved milestone

Deliverables considered complete only after I review and sign off on the milestone build.

When you reply, please include:
1. Links to one or two similar OCR or document-AI projects you have shipped.
2. The tech stack you propose and why it fits.
3. A fixed-price cost broken down by milestone with an estimated timeline.

I release payments strictly per finished milestone, so your plan should map naturally to the features above. Looking forward to seeing how you can help me move this MVP from idea to working prototype.