US Job Market Visualizer

This is an interactive tool for exploring real data about American jobs. Instead of reading tables, you see a colorful treemap — a grid of rectangles where each box represents a job category. The bigger the box, the more people work in that job. You can color the boxes by different metrics: how fast the BLS (government) expects that job to grow, median salary, education requirements, or estimated exposure to AI. It's a way to see patterns in the entire US job market at a glance.

The repo pulls all 342 occupations from the Bureau of Labor Statistics Occupational Outlook Handbook, which is the official source for job data in America. It then runs that information through an AI model to estimate how much each job might be reshaped by current AI tools. A software developer scores 9 out of 10 for "AI exposure," not because the job is disappearing, but because AI is actively changing what developers do every day. A job that scores high might grow *more* because workers become more productive, or it might shift in nature. The tool doesn't predict job loss — it highlights which roles are being transformed.

Behind the scenes, the project is a data pipeline. It starts by scraping the BLS website to grab all the occupation pages, then converts that messy HTML into clean structured data. A script extracts key facts like salary and job count into a spreadsheet. Then an LLM reads each occupation's description and assigns an AI exposure score with reasoning. All of that gets packaged into a single JSON file that powers the interactive visualization on the website.

The project is designed for exploration and conversation. Someone might wonder "which fast-growing jobs require only a high school degree?" or "what occupations have high pay but low AI exposure?" You can toggle between different color schemes on the treemap to answer those questions visually. The README emphasizes this is a research tool and exploration toy, not an economic forecast — the AI scores are rough estimates meant to spark thinking, not definitive predictions.