What to work through
Work through a ChatGPT-style chat as a spoken system design interview. Start by clarifying the scope, sketch a working design and explain the trade-offs behind your choices.
- Send a message and see the reply appear word by word
- Stop a reply while it is still arriving
- Retry a reply that failed or was stopped
- Come back to past conversations from a list
- Show formatted replies with lists and code blocks
Constraints to discuss
- The first words of a reply show within 1 to 2 seconds
- Typing and scrolling answer within 200 milliseconds while a reply streams
- Model output is treated as untrusted text before it is drawn
- Usable with a keyboard, with new replies announced to screen readers
Need a starting point?
Watch for these common gaps as you prepare:
- Re-formatting the whole reply on every new piece of text
- Sanitising each piece on its own instead of all the text so far
- A stop button that closes the connection while the server keeps generating
- No end state for a stream that just stops, so the reply spins forever
- Spending the round on server scale when the interviewer ruled it out
- Showing the server's raw error to the person instead of a short message and a retry
Why this question is in the collection
These sources connect the problem to company interview reports or preparation guides. They do not guarantee what a company will ask you.
- OpenAI: A software engineer reports an onsite round in July 2026, published October 2026: "design ChatGPT", with scale and saving set aside; then why one way of streaming over the other, and how long conversations still render well.
- OpenAI: A frontend loop tip from July 2025: the system design round was an older version of ChatGPT.
- OpenAI: A frontend engineer reports a phone screen, about February 2026: an AI chatbot UI with streamed replies and nothing stored on the server; follow-ups on the store's shape, one stream transport over another, browser storage limits and interrupted streams.
- Harvey: An undated tip on a full-stack round 2: a ChatGPT-style interface with document uploads and connected outside data sources; the candidate covered component structure, state and API design.
- Google: Listed by PracHub in April 2026 as a frontend engineer onsite question, an embeddable chat widget with streamed replies, with no candidate narrative.
- Waymo: Listed by PracHub in September 2026 as a frontend engineer onsite question, the frontend of an AI chat app with streamed replies, with no candidate narrative.