Harvey Interview Questions — Page 1

Harvey·Software Engineer·Technical Phone Screen

Apr 2026

HarveyAI software engineering interview centered on a file storage system design problem, with a pretty interesting follow-up that pushed into distributed systems territory. The core problem felt manageable but the second part is where things got real.

Design a file storage system ('The Vault') that supports adding files, reading files, deleting files, and listing files under a given directory path.
Given two separate instances of the vault (two replicas), design an efficient strategy to determine whether their file contents are consistent, minimizing the amount of data you need to transfer.

“I started with a trie-like structure for the directory tree and a hash map for file contents.”

View Post

Harvey·Machine Learning Engineer·Technical Phone Screen

Apr 2026

ML engineer interview at Harvey AI that centered on a technical deep-dive into a past AI project. Pretty much the whole conversation was about one thing: walk us through something you built, then defend every choice you made.

Walk us through an AI-focused project you've worked on previously.
What alternative approaches did you consider, and why did you go with the technologies you chose?
What were the pros and cons of the approach you took, and how well did it actually solve the problem?

“I had something prepped but the follow-up questions are where it gets uncomfortable.”

View Post

Harvey·Software Engineer·Technical Phone Screen

Apr 2026

Coding round at HarveyAI for a software engineer role. One question, pretty focused: build a connection pool with acquire and release, db.query was already mocked so you just had to get the pooling logic right.

Implement a database connection pool with acquire() and release() methods. db.query is pre-mocked; focus only on the pool management logic.

“Sounds straightforward until you actually think about concurrency.”

View Post

Harvey·Software Engineer·Technical Phone Screen

Apr 2026

HarveyAI coding round, got a file system design problem that looked deceptively clean until I started actually implementing it. Solid problem, probably worth grinding the path-parsing stuff before you go in.

Design and implement an in-memory file system that supports ls(path), mkdir(path), addContentToFile(path, content), and readContentFromFile(path).

“I went straight for a trie-style node structure where each node holds a name, a map of children, and a string for file content.”

View Post

Harvey·Software Engineer·Onsite - System Design / Architecture

Apr 2026

System design round at HarveyAI for a software engineer role. The interviewer pushed way deeper than I expected on two specific areas, file uploads and access control, to the point where it felt less like a breadth check and more like a technical interrogation.

Design the file upload flow for a cloud storage system like Google Drive, covering chunking, resumable uploads, deduplication, and how you separate metadata storage from blob storage while keeping them consistent.
How would you design the access control system for a cloud storage product, including folder permission inheritance, link-based sharing with different permission levels, and making ACL checks fast at read time?

“This is where the interview got uncomfortable.”

View Post

Harvey·Software Engineer·Technical Phone Screen

Apr 2026

Interviewed for a software engineer role at HarveyAI and got asked to walk through a major project in real depth, not just the happy path but the actual decision-making and trade-offs. It was one of those conversations where you realize mid-answer that you should've prepared a cleaner example.

Walk me through a significant project you worked on. What trade-offs did you consider, what alternatives did you evaluate, and why did you make the choices you did? What would you do differently now?

“I picked a project I knew well but kind of rambled through the alternatives part.”

View Post

Harvey·Software Engineer·Onsite - System Design / Architecture

Apr 2026

System design round at HarveyAI for a software engineer role. One question, pretty focused, and the interviewer was encouraging the whole way through which honestly made it easier to think out loud.

Design a file storage system like Google Drive. How would you handle very large files that need to be split into chunks?

“The core of it was separating metadata from actual file content.”

View Post

Harvey·Software Engineer·Onsite - System Design / Architecture

Apr 2026

System design round at HarveyAI focused almost entirely on building a Google Drive-style service, but the real meat was the access control layer. Dense interview, lots of follow-ups, left feeling like I covered maybe 70% of what they wanted.

Design the file and folder data model for a cloud storage service, including how parent-child relationships and permission inheritance work.
Walk through the sharing model: different permission levels like owner, editor, commenter, and viewer, and how sharing works across users, groups, links, and domains.
How do you handle permission propagation when a parent folder's ACL changes, especially when a child has an explicit override?
How would you evaluate access control checks at low latency and scale? What does a denormalized permission cache or a relationship-tuple-based system look like here?
How would you design the audit logging system for file access and permission changes?
Briefly cover how you'd handle file storage, client sync, and conflict resolution.

“I went straight to a tree structure with a parent_id foreign key and spent too long on the basic hierarchy before they nudged me toward permissions.”

View Post

Harvey·Software Engineer·Onsite - System Design / Architecture

Apr 2026

System design interview at Harvey for a software engineer role, focused entirely on building a RAG-based Q&A agent for a big law firm's memo corpus. It was a long open-ended design discussion with multiple layers, and the interviewer clearly wanted breadth across ingestion, retrieval, grounding, and evaluation rather than a deep dive into any single piece.

Design an AI system that lets attorneys ask natural-language questions and get answers grounded in a large corpus of legal memos, covering ingestion, retrieval, answer generation, and evaluation.
Your crawled corpus has two memos from different firms that directly contradict each other on the same legal question. How should the agent respond, and how do you surface that conflict to the user?
A regulator publishes a change that supersedes the conclusions in several memos already in your index. How does your pipeline prevent the agent from citing those stale memos as current authority?
Attorneys say the agent refuses too often on questions it could partially answer. How do you tune the refusal threshold without increasing hallucinations, and how do you measure that trade-off?
How would your design change if query volume grew 100x, or if you needed to support multi-turn conversations where follow-up questions depend on prior context?

“This was the whole interview basically.”

View Post

Harvey·Software Engineer·Technical Phone Screen

Apr 2026

Coding round at HarveyAI for a software engineer role. The problem was a spreadsheet cell dependency thing, basically a variant of a well-known design question, and I almost had it but ran out of time one line short.

Design a spreadsheet where cells can reference other cells, with the constraint that read operations must be O(1).

“One line away.”

View Post