Wukala-GPT
Document-grounded legal assistance: retrieval, agent workflows and an AI service connected to a full-stack application.
Useful context, usable product.
Legal document questions need relevant context and a usable path through the application. Wukala-GPT connects document analysis and retrieval with legal assistance and lawyer discovery.
Where the parts connect.
Read the flow as text
- React interface → request → Application APIs + auth
- Application APIs + auth → AI service → FastAPI + agent workflows
- FastAPI + agent workflows → retrieve → Documents + vector search
- Documents + vector search → context → Document-grounded response
Conceptual architecture based on the documented stack.
Clear boundaries between layers.
Retrieval and vector search provide document context for LLM-assisted responses. The AI capability sits behind an API boundary rather than directly inside the interface.
React, ASP.NET Core and FastAPI cover the interface, application services and AI service. PostgreSQL, Qdrant and Redis are part of the listed stack.
LangGraph supports the agent workflow layer; Docker is part of the development stack. Keep document context and application permissions aligned when designing this kind of system.
What is available today.
The project brings RAG, vector search, document analysis and LLM integration into one application. The available project record describes implementation and testing; it does not include public benchmark results.
A public demo, product screenshots and independent performance measurements are not currently published.
View Resume ↗