Online art galleries are great for humans and a nightmare for machines 😅: every artwork page is a bespoke little chaos of titles, prices and wildly inconsistent size strings. In this talk, we’ll take a real gallery (Artmajeur) and turn it into a clean, analytics-ready dataset — step by step — with modern Python tooling. 🎨🐍📊

We’ll start from an empty repo 🧱 and design a production-minded scraper: picking a modern scraping framework ⚙️, wiring up a single spider 🕷️, and hunting in DevTools 🔍 for the selectors that lead us from listing pages to artwork detail pages. From there, we’ll shape a clear data model — artist, title, price, raw size text, image URL — and build a small but realistic pipeline around it.

Then comes the fun part 🤹‍♀️: we’ll plug in a spaCy-powered NLP layer 🧠 to tame messy dimension strings and normalize them into consistent width_cm, height_cm, depth_cm, orientation, and size buckets — ready for SQL and dashboards.

You’ll walk away with a practical blueprint 🗺️ for going from “pretty website I’d like to scrape” to a robust, Picasso-Grade Python data pipeline that your future self (and your analysts) will quietly thank you for. 💫