reasearchSync

researchSync offers two services:

1. FastAPI-MCP Server: Connects to MCP Hosts like Claude Desktop to retrieve papers' metadata from arXiv and store it in a Notion database. You can then query this database to find the best paper you need.

2. Web App: Powered by Gradio (with plans to migrate to React in the future) for the frontend and FastAPI for the backend. It uses a Notion and Redis database to store metadata of the papers from arXiv and vector embeddings for retrieving the best papers based on user queries.

Python Version

• Python: 3.11

System Overview

The system will:

• Monitor arXiv publications from ML/AI related tags (e.g., 'cs.AI').

• Fetch metadata (title, authors, abstract) and add to Notion Database using Prefect workflow.

• Store data in a Notion database.

• Convert the paper's abstract to vector embeddings for fast retrieval and store on Redis.

• Allow an AI assistant to query or manage the database via MCP.

Prefect Workflow Snippet