Part 9 of 13 · 1 min read
Answering open-source questions in an interview
Tell me about an open-source contribution you made.
"I picked an agent library I use, reproduced a bug where a tool-call parser failed on a fenced JSON block followed by prose, wrote a test that failed on main, fixed it with a regular expression, linked the issue and explained how I tested. The change was about twenty lines and I responded to review comments." Follow-up: "What would you do differently?" Name one thing, such as checking for existing issues or running the project's linter first. (If you have not contributed yet, say so and describe the exercise honestly.)
How do you make a pull request easy to review?
"One concern, a small diff, a test that fails before and passes after, a description with the issue link, the reason and how I tested, no unrelated formatting, and my own review of the diff before asking for anyone else's." Follow-up: "A reviewer asks for a bigger refactor." Offer a follow-up pull request and keep this one focused.
How do you choose a project and an issue?
"A project I use, with recent activity, responsive maintainers and a contributing guide; an issue I can reproduce, with a clear scope. I start with docs, tests or a small bug before features." Follow-up: "How do you know the project is active?" Release dates, merged-pull-request age, issue response time.
What do you check about licences?
"That the project's licence permits my use and that my contribution is licensed as the project requires; for models, whether the licence covers commercial use and what data and weights are available." Follow-up: "Can you paste code from a Stack Overflow answer or an AI tool?" Only if you understand it, it is compatible with the licence and the project's policy allows it.