National Parks × Chalk: A DevRel Take-Home, Built and Launched
As a DevRel, you’re expected to do it all: build the thing, write about it, and figure out how to get it in front of the right people. This project is that workflow end to end: a technical take-home I built, documented, and then planned distribution for, treating it like a real content launch instead of a code submission.
The problem
Starting from a basic Chalk project and the National Park Service API, the question was: how many hours will a sales rep spend driving between the national parks in their state? Chalk is a feature engineering platform: you declare what you want to know about your data, and it works out how to compute and serve it.
What I built
- Enriched the data with each park’s activities and GPS coordinates.
- Filtered for relevance with resolvers like
has_fishingandhas_water. - Calculated distances for every pair of parks in a state with the Haversine formula: 378 pairs for California’s 28 parks.
- Optimized the route with a nearest-neighbor heuristic, plus a per-park version to compare starting points. Starting from Yosemite saves about 3 hours over Alcatraz.
How I’d launch it
One project, five assets, each for a different stage of the developer journey: the blog post, the README, a 2-minute video built with code (Remotion, with narration generated by Groq’s text-to-speech), the raw working notes, and the runnable code itself. Building the video in code means a teammate can review it in a pull request and regenerate it when the product changes.