OpenAI Interview Questions — Page 64

OpenAI·Software Engineer·Onsite - System Design / Architecture

Jul 2026

OpenAI onsite system design round for a software engineering role. The session wrapped up early because the interviewer ran out of follow-ups, which honestly felt like a good sign, and the two of us just ended up chatting for the rest of the time.

Design a video generation system similar to Sora.

“Wasn't expecting to finish early but we burned through all the follow-ups and the interviewer just kind of shrugged and said there was nothing left to dig into.”

View Post

OpenAI·Software Engineer·Onsite - Coding / Algorithms

Jul 2026

Coding round at OpenAI for a software engineer role, focused entirely on designing a rate limiter with distributed and persistence requirements. The interviewer pushed hard on failure scenarios, which is where things got interesting.

Design a rate limiter that works in a distributed environment and supports persistence. Implement the core request-handling and reset logic.
How would you handle clock skew across distributed servers in your rate limiter?
What happens if Redis goes down? How do you fall back to local storage without breaking availability?

“Started by talking through the approach before writing anything, which I think helped.”

View Post

OpenAI·Software Engineer·Onsite - System Design / Architecture

Jul 2026

System design round at OpenAI for a software engineering role, one massive question covering basically every dimension of a payment platform. The scope was genuinely intimidating and I don't think I covered everything they wanted.

Design a scalable payment system that supports peer-to-peer transfers, merchant card and digital wallet payments, refunds, chargebacks, and full transaction history, including integration with external payment rails.

“This question ate the whole session.”

View Post

OpenAI·Software Engineer·Onsite - System Design / Architecture

Jul 2026

System design round at OpenAI for a software engineering role, focused entirely on building a real-time streaming chat UI in React. The question was deceptively deep and kept branching into concurrency, state management, and reliability edge cases I wasn't fully prepared for.

Walk through how you'd design and build a React chat interface where assistant responses stream in token by token, users can send new messages mid-stream, and multiple conversation threads can be active at once.
Name two pieces of state that only exist while a stream is active and explain how and when you clean them up.
How do you make sure streamed tokens append correctly without flickering or overwriting content, especially when state updates are happening rapidly?
What can go wrong when multiple responses are streaming concurrently, and how do you prevent one stream from corrupting another's state?
When would you use useRef versus useState in this streaming scenario, and what's your general rule for deciding?
If the same network request gets sent twice due to a double-submit, retry, or StrictMode remount, how do you prevent a duplicate assistant message from appearing?
React 18 StrictMode mounts effects twice in development. How does that interact with opening a stream inside useEffect, and how do you make the setup and teardown idempotent?
The connection drops halfway through a streaming response. How do you handle recovery, and what do you need from the server contract to support it?
Rendering on every token tanks performance for long message lists. How would you diagnose this and what techniques would you apply?
If you moved the entire streaming state machine out of React component state, where would it live and what do you gain or lose compared to keeping it in useReducer?

“This started as a React question and turned into a full system design conversation.”

View Post

OpenAI·Software Engineer·Technical Phone Screen

Jul 2026

Applied stats heavy technical screen for a research engineer role at OpenAI. The whole thing was basically one long derivation problem about signal detection across GPU nodes, and it went deeper than I expected.

You have multiple GPU nodes where each node outputs a signal corrupted by Gaussian noise. Walk through how you'd set up the statistical model, derive a test for whether a true signal is present, combine observations across nodes optimally, choose a test statistic, set a detection threshold for a given false-positive rate, and explain how power scales with the number of nodes.

“This was basically one question that kept unfolding for 40 minutes.”

View Post

OpenAI·Machine Learning Engineer·Technical Phone Screen

Jul 2026

Interviewed for an ML Engineer role at OpenAI and got hit with a systems-flavored coding problem that I wasn't expecting to go as deep as it did. The dependency resolution angle made it feel more like a compiler or package manager design question than a pure algo problem.

Given a list of packages where each package has a set of dependencies (potentially with version ranges), write an algorithm to find a valid installation order or determine that no valid order exists. Be prepared to discuss how you'd handle dependency cycles, diamond dependencies, and version constraint conflicts.

“I started with topological sort which felt right, but then they kept pushing on version ranges and I realized I'd basically been ignoring half the problem.”

View Post

OpenAI·Machine Learning Engineer·Onsite - Coding / Algorithms

Jul 2026

ML coding round at OpenAI for an MLE role. The whole thing centered on one meaty calibration problem and then a discussion portion that went places I wasn't fully ready for.

Implement a calibration utility for LLM outputs: given logit/probability and ground-truth label pairs from a held-out set, fit a calibration map using temperature scaling, Platt scaling, or isotonic regression so that reported confidence matches empirical accuracy. Then write a function that takes a new logit vector and returns a calibrated probability distribution.
Walk through how you'd construct a reliability diagram and compute expected calibration error (ECE), and discuss the trade-offs between temperature scaling, Platt scaling, and isotonic regression as calibration methods.

“I went straight to temperature scaling because it's the simplest and I could actually derive the optimization cleanly under time pressure.”

View Post

OpenAI·Machine Learning Engineer·Onsite - System Design / Architecture

Jul 2026

System design round at OpenAI for an ML Engineer role, focused entirely on building a RAG-based search and QA system over a large mixed-format document corpus. The whole session was basically one giant design problem with follow-ups layered on top, which felt more like a technical deep-dive than a typical interview.

Design an LLM-powered search and question-answering system over a large corpus of documents including internal wikis, PDFs, logs, contracts, and web pages. Both documents and user queries can exceed the model's context window.
How do you handle documents that are hundreds of pages long and far exceed what you can fit into the context window?
How would you process a long user query, like a multi-page pasted log or contract, without losing the user's original intent?
How would you evaluate whether your chunking strategy is causing missed answers?
How would you handle questions that require synthesizing information across many documents rather than finding a single fact?
How do you handle permissioned documents in retrieval and caching without leaking data across users or teams?
What would change in your design if the task were summarizing an entire 200-page document rather than answering search queries over a corpus?

“This is the main question and it's a lot.”

View Post

OpenAI·Machine Learning Engineer·Onsite - System Design / Architecture

Jul 2026

System design round at OpenAI for an MLE role. The whole thing was a single deep-dive into building a production RAG system end-to-end, with follow-ups that kept branching into trickier territory. Felt more like a 45-minute architecture review than a standard interview.

Design the end-to-end architecture of a RAG system for question answering over a large internal document corpus, covering ingestion, chunking, embedding, retrieval, reranking, and generation with citations.
Why does pure dense vector retrieval fall short for certain query types, and how would you combine it with lexical search to improve retrieval quality?
How would you reduce hallucinations and ensure the generated answers are grounded in the retrieved source documents?
How would you handle documents with strict access controls so users can't see results from documents they're not permitted to access?
Walk through how you'd debug a case where the system confidently cited the wrong document. How do you determine whether the failure is in retrieval or generation?
Your embedding model gets upgraded to a new version. How do you re-index without creating a broken mixed-embedding-space index during the transition?
How would you extend the system to answer multi-hop questions that require combining facts from multiple documents?
The corpus has many near-duplicate documents. How does that affect retrieval quality and what would you do about it?
Describe your evaluation plan for a RAG system, covering both offline metrics and online monitoring, including how you'd detect corpus or embedding drift over time.

“This was the core question and it ate up most of the session.”

View Post

OpenAI·Software Engineer·Technical Phone Screen

Jul 2026

Got a system design question at OpenAI for a software engineer role, focused on data structure choices for a social graph lookup. Pretty focused interview, just the one problem but they really wanted to dig into the tradeoffs.

You're given a static snapshot of a social network's follow relationships. Design a method `follows(a, b)` that checks whether user A follows user B. What data structure would you use to store the snapshot, and what query complexity would you expect?

“I went with a hash map of sets, one entry per user, each set holding the accounts they follow.”

View Post