Case 07In the open00:00:07:00
← All workNexusAI
00:00:00:00Play
- Scope
- Open-source build
- Core stack
- FastAPI · React · TypeScript
- Role
- Architecture, build, deploy
The outcome
Local-first RAG system for handwritten notes: ingests PDFs page by page - text, diagrams, tables, equations, and margin notes - into a local store, then answers questions through a streaming React chat interface.
The build
Every answer cites the exact PDF and page it came from, never model memory.
Blueprint — how NexusAI fits together
6 parts · 5 connections
Source
Ingest
Store
Answer
Every answer cites the exact page it came from.
Hover, focus or tap a part of NexusAI to trace its connections.
- PDFs connects to Page-by-page ingest
- Page-by-page ingest connects to Local store
- Local store connects to FastAPI
- FastAPI connects to React chat
- FastAPI connects to Citations
Credits — the stack
- FastAPI01
- React02
- TypeScript03
- RAG04
- SSE05
- Local-first06
- jsPDF07
