The application allows users to upload PDF documents, ask questions about their content, and receive AI-generated answers with relevant document sources.
Here's a quick look at the application in action!
**Technology Stack**
Backend:
- ASP.NET Core Web API (.NET 9)
- Entity Framework Core
- Redis
- RabbitMQ & MassTransit for asynchronous processing
-Cloud-based PDF storage using Cloudflare R2
Frontend:
- Angular 22
- TypeScript, HTML, and SCSS
AI & RAG
- Retrieval-Augmented Generation (RAG)
- PDF text extraction and chunking
- `multilingual-e5-small` for text embeddings
- Qdrant for vector search
- Ollama with Llama 3.2 for AI-generated responses
Authentication & Security:
- Email verification and password recovery
- Google authentication integration
- Backend-for-Frontend (BFF) session management
- Protected Angular routes and role-aware authentication
Key Features:
- PDF upload and document processing status tracking
- Asynchronous processing with retry support
- AI-powered conversations with document source references
- Server-side pagination, filtering, and sorting
- User-specific document access
What I Focused On :
My goal was to go beyond basic CRUD operations and build an application that brings together backend engineering, asynchronous workflows, secure authentication, vector search, and LLM-powered question answering.
Working on KnowledgeHubAI has helped me deepen my understanding of how these technologies fit together to build practical AI-powered applications.