Project Overview

This project demonstrates the engineering of a specialized web application designed to automate the extraction and structuring of data from PDF documents concerning German reforms. It processes documents written in either German or English, consistently delivering structured insights in English. The platform integrates AI for sophisticated text analysis, employs asynchronous processing for performance, and utilizes modern DevOps practices for reliable deployment, showcasing the ability to build robust, data-centric solutions.

Key Technologies

• Backend: ASP.NET Core, C#, Quartz.NET, SignalR, FluentValidation
• Frontend: React, TypeScript, React Query, React Router
• AI Integration: Google Gemini 2.0 Flash API
• Database: PostgreSQL with Entity Framework Core (EF Core)
• Infrastructure & DevOps: Docker, Kubernetes (K8s), GitHub Actions (CI/CD)

Backend Architecture and Implementation

The backend leverages ASP.NET Core combined with Minimal APIs to create efficient, lightweight endpoints suitable for a focused application. A pragmatic Layered Architecture was implemented to maintain clear separation between presentation, application logic, and infrastructure concerns, balancing structure with development velocity for this specific use case.

A core challenge was handling the potentially time-consuming AI analysis. This was addressed using Quartz.NET to manage asynchronous background jobs. Upon PDF upload, tasks are queued, decoupling heavy processing from the user request and ensuring UI responsiveness. Quartz.NET provides reliability for job execution.

To perform its core analytical task, the platform integrates with Google Gemini 2.0 Flash. To accurately extract approximately 20 distinct data fields from the document text, I developed and utilize a strategy involving multiple, parallel-executed, specialized prompts. I also implemented a retry mechanism that handles potential AI response formatting issues, enhancing reliability. Once the analysis is complete, SignalR delivers real-time feedback directly to the user's browser, eliminating the need for manual page refreshes. Server-side validation using FluentValidation ensures that invatiants are kept intact after the document submission.

Persistent data storage is handled by PostgreSQL, accessed efficiently through Entity Framework Core (EF Core).

Frontend User Interface

The UI is constructed using React and TypeScript, providing a type-safe and component-based structure. A key element is the use of React Query for managing server state. This library significantly simplifies data fetching, caching, and synchronization related to the backend processing status and results. It handles loading/error states gracefully and efficiently updates the UI when new data arrives via the SignalR connection.

The user experience is designed for simplicity: upload a PDF, receive immediate feedback that processing has started, and see structured results appear dynamically. React Router manages navigation. Client-side validation provides quick feedback, complementing the backend rules. The interface ensures user context (the document being processed) is maintained even if the page is reloaded.

Deployment Automation and Infrastructure

Modern DevOps practices ensure automated and consistent deployments. A CI/CD pipeline configured in GitHub Actions automates the entire workflow: building the backend and frontend, packaging them into Docker containers, and deploying these containers to a Kubernetes (K8s) cluster. This containerization ensures environmental consistency, while K8s manages application scaling, deployment rollouts, and resilience.

Conclusion

This Reform Proposal Hub project demonstrates practical expertise in building specialized, data-focused web applications. It effectively integrates AI for complex data extraction, utilizes asynchronous patterns (Quartz.NET) and real-time communication (SignalR) for a responsive user experience, and employs modern frontend state management techniques (React Query). The implementation showcases strong skills across the stack (ASP.NET Core, React, TypeScript, PostgreSQL) and proficiency in contemporary DevOps workflows (Docker, Kubernetes, CI/CD), representing the capability to engineer and deploy robust, automated solutions.