OpenAI Interview Questions — Page 25

OpenAI·Software Engineer·Technical Phone Screen

May 2026

Got a coding question at OpenAI that was basically a twist on the classic rotting oranges problem, except with a threshold mechanic that changes everything about how you approach the BFS. Took me a bit to realize why my standard solution wasn't going to cut it.

Given an m x n grid where cells are either empty, healthy, or infected, simulate infection spreading each minute. The twist: a healthy cell only becomes infected if the number of currently infected neighbors is at or above a given threshold k. Return the number of minutes until no more healthy cells can be infected, or -1 if some healthy cell will never be infected.

“My first instinct was to just run BFS like the standard version of this problem and I started coding that up before I caught myself.”

View Post

OpenAI·Software Engineer·Technical Phone Screen

May 2026

Technical screen for a Research Engineer role at OpenAI. One meaty ML question about annotator quality filtering and iterative retraining. The kind of problem that sounds scoped until you start thinking about edge cases.

You have a classification dataset with human-provided labels and per-annotator IDs. Some annotators are noisy. Walk through how you'd score each annotator's reliability, filter out low-quality labels, retrain the classifier, and confirm whether validation accuracy actually improves. Also address how you'd pick thresholds, avoid label leakage, and handle annotators who only have a few labels.

“This one took me longer to get into than I expected.”

View Post

OpenAI·Software Engineer·Technical Phone Screen

May 2026

OpenAI interview that went deep into compiler/type-system territory. One question, but it was a beast. The kind of problem where you think you understand it after five minutes and then realize you've barely scratched the surface.

Design and implement a type system for a small toy language. It should support primitive types, generics, nested tuples, and function signatures, including a type inference engine that resolves generics to concrete types and handles mismatches or conflicts.

“I started with the data structures and felt okay there, primitives and tuples weren't bad.”

View Post

OpenAI·Software Engineer·Onsite - System Design / Architecture

May 2026

System design round at OpenAI for a software engineer role, 60 minutes, focused entirely on payment processing. The interviewer was thorough and the topic had enough depth to fill the whole session without any fluff.

Design a payment processing system.

“This is the kind of question where you can go in a dozen directions and pick the wrong one fast.”

View Post

OpenAI·Software Engineer·Onsite - System Design / Architecture

May 2026

System design round at OpenAI for a software engineering role, focused entirely on designing a GPU job scheduler for a text-to-video generation service. Pretty deep dive, multiple parts, and the follow-ups got increasingly brutal toward the end.

Walk through the full request lifecycle for a long-running GPU generation job: how does a submission get accepted, where does it wait, how does it get dispatched to a worker, and how does the client get status and results back?
How do you design the scheduling policy across multiple user tiers with different SLOs, while making sure no tier gets completely starved under sustained load?
How do you enforce per-tier and per-customer quotas at admission so the scheduler isn't overwhelmed trying to be fair across traffic it should have rejected upstream?
When a high-priority job arrives and all GPUs are occupied, how do you decide which running job to preempt, and do you checkpoint it or just kill and restart?
How do you allocate GPUs to multi-GPU jobs without causing fragmentation or leaving GPUs idle while a job waits for the rest of its allocation?
One enterprise customer suddenly submits 10,000 jobs at once. How do you protect other enterprise customers and lower tiers, and what does the offending customer actually experience?
If you wanted to add GPU fleet autoscaling, how does the multi-minute provisioning lag change your queuing and preemption strategy?
Your fleet has multiple GPU SKUs and some jobs only run on certain hardware. How does the scheduler handle job-to-hardware compatibility?
Identical prompts and parameters come in repeatedly. How would you add result caching or deduplication, and what are the correctness and privacy risks?
How do you keep the scheduler itself highly available, and what prevents a job from being dispatched to two workers simultaneously after a scheduler crash?

“This is the foundation question and I spent probably too long here.”

View Post

OpenAI·Machine Learning Engineer·Technical Phone Screen

May 2026

Coding round at OpenAI for an ML Engineer role, centered entirely on implementing type-checking logic. Pretty niche focus but it tested recursion and edge case thinking more than I expected.

Implement a type-checking routine for a data structure specified by the interviewer, such as validating that a dict or JSON object conforms to a declared schema with required keys, expected types per field, and nested objects.

“The recursive part wasn't hard to sketch out, but I kept second-guessing myself on the edge cases.”

View Post

OpenAI·Software Engineer·Technical Phone Screen

May 2026

ML coding round at OpenAI for a SWE role, focused entirely on cleaning human annotation data. The problem was open-ended in a way I wasn't fully prepared for, more of a system design meets ML pipeline question than a typical coding problem.

Design and implement a method to detect and filter bad annotations from human-labeled data. Given labeled samples with optional annotator IDs, confidence scores, timestamps, and a small gold set, determine which annotations to drop and which annotators to downweight.

“This one sprawled in a way I didn't expect.”

View Post

OpenAI·Software Engineer·Onsite - System Design / Architecture

May 2026

System design round at OpenAI for a software engineer role. The question was a single massive prompt covering basically every dimension of a ChatGPT-like service with preset/system-prompt management. Felt like they wanted to see how you'd prioritize under pressure rather than actually finish the whole thing.

Design a minimal ChatGPT-style conversational service that supports reusable system prompt presets per user and team. Cover core components, data model, APIs, session/context management, storage, scalability, multi-tenancy, safety, failure handling, extensibility, and capacity estimates for 100k DAU. Also walk through the call sequence for creating a preset and sending a message using one.

“This is basically five system design questions duct-taped together.”

View Post

OpenAI·Software Engineer·Onsite - Coding / Algorithms

May 2026

Coding exercise at OpenAI for a software engineer role, 60 minutes to design and implement a turn-based two-player game simulation from scratch. The scope was broader than I expected going in.

Design and implement a two-player turn-based game simulation (like tic-tac-toe) that models game state, handles move validation, detects win/loss/draw conditions, and can predict the winner from any mid-game state. The design should be extensible to other similar games.

“60 minutes felt fine until I got to the winner prediction part and realized I'd painted myself into a corner with my class structure.”

View Post

OpenAI·Software Engineer·Onsite - Behavioral / Leadership

May 2026

Interviewed for a growth marketing role at OpenAI, behavioral round with at least one question digging into experimentation and failure.

Walk me through an experiment you ran that didn't work out.

“I had a story ready but fumbled the landing.”

View Post