OpenAI interview notes, page 21 of 67

Mid-sized

Most of these OpenAI write-ups come from Software Engineers (460), then Machine Learning Engineers (137) and Backend Engineers (23). The most reported rounds are the technical phone screen (335), the system design round (229), the onsite coding round (32) and the behavioral / leadership round (22).

667 interview notes · updated Jul 2026

OpenAI·Software Engineer·Onsite - System Design / Architecture

May 2026

System design round at OpenAI for a software engineer role. One meaty question about building a key-value store with crash recovery, which sounds straightforward until you actually have to talk through all the moving parts under pressure.

Design an in-memory key-value store with get, put, and delete operations. Every write must be persisted to a write-ahead log so the store can recover its state after a crash or restart. Walk through your log format, how you handle appends, fsync behavior, and optionally how you'd compact the log over time.

“I started with the happy path, get/put/delete backed by a hashmap, and that part was fine.”

View Post

OpenAI·Software Engineer·Technical Phone Screen

May 2026

Got a coding round for a Software Engineer role at OpenAI. The whole thing was one fairly involved design-and-implement problem about a credit ledger, with a bunch of test cases and a complexity discussion tacked on at the end. More work than I expected for a single question.

Design and implement an expiring GPU credits ledger that supports multiple users. It needs three operations: adding a credit lot with an expiry time, subtracting credits atomically by consuming lots in earliest-expiry-first order (failing entirely if balance is insufficient), and querying the current balance with a per-lot breakdown. Lots expire when timestamp >= expiry_time.
After implementing the ledger, analyze the time and space complexity of each operation, and discuss how you'd optimize for a high volume of operations. How would you achieve O(log n) per update, and how would you handle persisting and restoring the ledger state?

“This took me a minute to fully parse.”

View Post

OpenAI·Machine Learning Engineer·Onsite - Behavioral / Leadership

May 2026

Went through a Team Chats round at OpenAI for an MLE role, which was basically behavioral. Focused on collaboration, cross-functional work, and how you handle friction with other teams.

How do you collaborate with others on your team, and what does that look like day to day?
Tell me about a time you ran into a significant difficulty at work. How did you handle it?
How do you work with teams outside your immediate function, like product or design, to get things done?

“Pretty standard stuff but I fumbled it a bit by jumping straight into a project story without setting any context.”

View Post

OpenAI·Software Engineer·Onsite - System Design / Architecture

May 2026

OpenAI software engineer interview, system design round focused on low-level memory management. The problem had some depth to it and the follow-up discussion pushed into real tradeoff territory.

Design a memory allocator that supports coalescing of free blocks to reduce fragmentation.

“Went with a linked list of free blocks and handled merging adjacent blocks on deallocation.”

View Post

OpenAI·Software Engineer·Technical Phone Screen

May 2026

Interviewed for a SWE role at OpenAI, got a simulation-style coding problem involving infection spread. Not a lot to say since the details are fuzzy in my memory now, but it was a technical screen of some kind.

Simulate how an infection spreads through a population or grid over time.

“I remember it being a grid traversal type problem but I can't recall the exact constraints.”

View Post

OpenAI·Software Engineer·Onsite - System Design / Architecture

May 2026

OpenAI infrastructure round, one big system design question that spiraled into a pretty deep conversation about durability guarantees and crash recovery. Not a casual chat.

Design an in-memory key-value store that can recover its full state after a crash or disconnect, using a write-ahead log. Cover the on-disk log format, durability semantics for put and delete operations, replay logic on startup, and optionally snapshotting with log truncation. How do you handle partial or torn writes?

“I started with the log format and that went okay, length-prefixed records with a checksum at the end, pretty standard stuff.”

View Post

OpenAI·Machine Learning Engineer·Technical Phone Screen

May 2026

OpenAI MLP engineer interview where they had me implement backprop entirely from scratch, no autograd crutch. Pretty intense for a technical screen but felt very on-brand for them.

Implement backpropagation from scratch for a small multi-layer neural network, manually computing gradients for each layer using the chain rule, without using any automatic differentiation library.

“This one took me a minute to settle into.”

View Post

OpenAI·AI Engineer·Technical Phone Screen

May 2026

Interviewed for an AI Engineer role at OpenAI and got hit with a pretty gnarly algorithmic scheduling problem. The whole thing revolved around generating balanced sequences of triples across multiple dimensions, which is not the kind of thing you can just wing with a standard LeetCode pattern.

You have three tag dimensions (AI, Human, Task) each with a finite set of values. Design an algorithm to generate a sequence of N triples, one from each dimension, such that at every prefix of the sequence, the count of any pair of values across dimensions stays as balanced as possible (max count minus min count is at most 1). Walk through your approach.

“This one took me a minute to even parse.”

View Post

OpenAI·Software Engineer·Technical Phone Screen

May 2026

Coding round at OpenAI for a software engineer role. One problem, a variation on the classic hit counter design, and I ran out of time before even seeing the follow-up. Not my best showing.

Design a chat tracking system with two methods: one to record a chat event given a user ID, chat ID, and timestamp, and another to return the count of chat events within the last 15 minutes for a given user and chat.

“Basically a hit counter problem but the window is anchored to the last recorded event, not the current time.”

View Post

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

May 2026

System design round at OpenAI for an MLE role. The format was a combined bug hunt plus system design session, which I wasn't fully expecting. You pick a non-trivial system you've actually worked on and walk through the whole thing: requirements, architecture, data model, scaling, trade-offs. Felt more like a conversation than a grilling, but that also meant I had to drive it myself, which is harder than it sounds.

Walk through a non-trivial system you've built or know well, covering requirements, high-level architecture, data model, scaling considerations, and the key trade-offs you made.

“The open-ended framing is deceptively hard.”

View Post