Cursor for Data Science

Isaac Steinberg

• MBA student
• President, Cornell Tech AI Society
• Software engineer with 7 years of professional experience in at startups, consulting
• Building applications for renewable energy companies and investment banks
• Coder and artist
• Interested in traveling, the natural world, and astronomy

Why Use Cursor?

• Fork of VS Code with AI-native features
• Agent, autocomplete, and Composer (what we'll use today)
• Writes and runs code in any language (Python, Node, etc.)
• Connects to external tools via MCP servers (browser, docs, APIs)

GUI vs Terminal AI Tools

Why Cursor over Claude Code or other terminal-based tools?

• See outputs visually — easier to read code
• Visual extensions — data viz, maps, charts
• Richer environment to interact with code
• Better for writing, reviewing, and understanding code

What We'll Cover Today

• ✓ Phase 0: Download Data — data/
• Phase 1: Text Overview
• Phase 2: Charts
• Phase 3: Interactive Website
• Phase 4: Experimental

The Datasets

Download pre-processed data

Phase 1: Text Overview

Summary statistics and initial observations

• Explore each dataset
• Identify patterns and anomalies
• Output: outputs/summary.md

Phase 2: Charts

5 visualizations to understand the data

• Hourly ridership patterns
• Income by subway line
• Bar density near stations
• Radar charts by line
• Output: outputs/charts/*.png

Phase 3: Interactive Website

HTML/JS dashboard

• Interactive map of stations
• Line explorer with demographics
• Output: website/index.html

Phase 4: Experimental

Creative data exploration

• Sonification — hear the data
• 3D visualization
• Flow animation
• Output: experimental/*

Let's Build!

Using Cursor to explore NYC subway demographics

Resources

Slides by Isaac Steinberg MBA '26