I had an idea the other day when working on some AI voice stuff... what if you could re-create the old birding hotlines but have them automatically update upon recent sightings? What would it be like to be able to call a number and hear which birds are currently being seen in an area, knowing that the information was always up to do? How feasible is this with current tooling and would it even be fun to use? Let's see!

In part 2 of this exploration we dug deep into how to answer the question "where do all the birders go?" We found many ways to make queries run very slowly on DuckDB but ended up hitting our initial goal of starting to use the EBD in production. Now, we're going to extend that further to see if we can show people how many birds might be likely in a given place and time.

In the last part we looked at the "basics" of getting DuckDB set up with the full eBird dataset. It was slow, it used a lot of data, but we finally got to answer the real questions, like "Who's seen the most ducks?" Now, it's on to slightly more practical explorations. The goal in this post is to go from a large, still relatively unwieldy dataset, to something we can actually deploy and host on the personal site

Ever since I started birding, I’ve had an ever growing list of questions. Questions about the species I’ve seen, the ones I haven’t, the habitats they’re found in, the activity and the participants itself. When do Eastern Phoebes typically show up in New York? Where do they go when they leave? Where do they like to nest? How frequently are they seen in New York? Why does that frequency dip in the summertime?

I rebuilt my personal site using Claude Code. Here's how we went from a Frankenstein React setup to a Docusaurus site, with Claude handling most of the implementation.