Twitch API Endpoint Walkthrough: Method, Parameters, Response and Auth for a PM
Company: Twitch
Role: Product Manager
Category: Software Engineering Fundamentals
Difficulty: medium
Interview Round: Other
Give one example of an API endpoint that Twitch could expose, and walk through it: what it does, how a client calls it, what comes back, and who would use it.
```hint Start from a user action
Pick something a viewer, streamer or outside developer does every day on Twitch, then work backward to the request that action needs and the data the response has to carry.
```
### Clarifying Questions
- Should the endpoint be one Twitch offers to outside developers, or one its own apps call behind the scenes?
- Do you want a REST-style endpoint, or is another style, such as GraphQL or an event subscription, fine?
- How deep should I go: just the method, path and response, or also authentication, pagination and rate limits?
### What a Strong Answer Covers
- A concrete endpoint tied to a real Twitch use case, with its HTTP method and resource path
- Inputs and output: the parameters, a sample response and why each field is there
- Authentication and permissions: whose data the call can read or change
- Operational concerns that matter for live video: paging through a fast-changing list, caching, rate limits and errors
- The product view: who calls the endpoint, what it enables for Twitch, and how its health and adoption would be measured
### Follow-up Questions
- A chat bot wants to post a message the moment a streamer goes live. Should it poll your endpoint or use something else, and why?
- How would you change the endpoint without breaking the third-party apps that already depend on it?
- What would you track to know whether the endpoint is healthy and useful to developers?
Overview: A technical product manager question asking the candidate to describe one API endpoint Twitch could offer. It tests whether a PM can explain an endpoint's method, path, parameters, response and authentication, reason about pagination, caching and rate limits for live data, and tie the endpoint to developer and product value.