Advanced Data Science
  • Home
  • Schedule/Syllabus
  • Homework and Presentations
  • Instructors
    • Brian Caffo
    • John Muschelli
  • Resources

On this page

  • Goal
  • Requirements

Homework 4

Pipeline with APIs

Author

Term 2 - HW 4

Goal

Create a reproducible, failure-aware pipeline that retrieves data from an API and runs downstream work only when the relevant data have changed.

Requirements

  • Choose a documented public API and identify the endpoint, license, update semantics, and rate limits.
  • Cache the raw response with metadata (retrieval time, URL/parameters, and content hash or stable version identifier).
  • Implement change detection that avoids unnecessary downstream computation while allowing a forced refresh.
  • Validate the schema and fail with a useful message if expected fields change or disappear.
  • Produce one derived artifact and a provenance record; test the unchanged, changed, and malformed-response paths.
  • Submit source code, a README.md, a small example cache/fixture where licensing permits, and a short architecture diagram.

AI may help draft client code, but you must validate it against the API documentation and include prompts plus tests in AI_USE.md. Do not put API secrets in the repository or an AI prompt.