Abhinav Gupta

Fetching the page is the easy part. Most of my time goes on what comes back.

I've lost hours to scrapers that changed shape when I toggled an option, to salary ranges I had to regex out of display text, and to a "remote" flag that turned out to include hybrid roles. Small things, but they're the ones that break a pipeline a week after you've forgotten how it works.

So my Actors are boring on purpose. Missing values come back as null rather than vanishing, so the columns are the same on every run. Fields are typed and described in the output schema, so you can see what you'll get before you spend anything on a run. Salaries get parsed into numbers with a currency instead of being left as text.

Where a site publishes an official API, I use it. That's less clever than working around bot detection, and it holds up better, which matters more to me than a scraper that impresses once and quietly dies.

If the output looks wrong, file an issue on the Actor and I'll fix the schema.