Coding Test Project 2: LLM and AI Agent

Description

Tools: LangChain and LangGraph, or other LLM frameworks such as AutoGen, Crew; The same full-stack development tools mentioned in Project 1; Vega-Lite: a grammar of interactive graphics (DO NOT use D3.js in this task) Vega-Lite online editor: https://vega.github.io/editor/#/examples/vega-lite/bar T1: Create a small sample of data from SciSciNet, for example, all papers from your current university (and filter other tables accordingly). T2: Build a multi-agent LLM framework to automate the data analysis process: data filtering, data analysis, and data visualization. Output: Perfect: A web-based chatbot with a chat interface and one or more dashboards Or you can also provide a Jupyter notebook to show the workflow. Coding Test 3 Interaction: User asks an example question such as: show me the number of papers by field or show me the number of papers by year . The system returns an interactive dashboard using Vega-Lite (e.g., a bar chart with filtering and hovering interactions)