Snowflake Interview Questions — Page 4

Snowflake·Software Engineer·Onsite - System Design / Architecture

Apr 2026

Snowflake system design interview that went deep into distributed systems, specifically around tree-based node counting using async message passing. The question had a lot of moving parts and I don't think I nailed all of them.

You have an N-ary tree where each node runs on a separate machine as a distributed component. Each node knows its parent and children. Implement a method that uses asynchronous message passing to return the total node count to the initiating node, using exactly two message types (REQUEST_COUNT and REPLY_COUNT). Cover the message format, per-node state, how requests propagate and replies aggregate, how you prevent duplicate propagation, how the initiator detects termination, and your concurrency/delivery assumptions plus complexity.

“This one had six sub-parts and I started rambling through them in order which was probably a mistake.”

View Post

Snowflake·Software Engineer·Onsite - System Design / Architecture

Apr 2026

Snowflake system design round, one massive question that basically covered every distributed systems topic I'd ever studied crammed into a single scenario. Walked out feeling like I'd either nailed it or completely bombed it, no idea which.

Design a resilient and secure end-to-end request flow where clients must first acquire a token from an unreliable third-party authorization service before calling your main API. Cover token acquisition, validation, caching and refresh, retries with exponential backoff and jitter, circuit breakers, request hedging, idempotency keys, fallback paths, poison-pill detection for malformed responses, observability, rate limiting, SLA/SLO definitions, and blast-radius containment across regions. Include API design, data models, and failure-mode playbooks.

“This question is basically a whole system design interview compressed into one prompt.”

View Post

Snowflake·Backend Engineer·Onsite - System Design / Architecture

Apr 2026

System design round at Snowflake for a backend role, focused entirely on building a cross-platform client logging library end to end. The interviewer kept drilling deeper on each sub-topic so it felt less like one question and more like six back-to-back. Solid round if you like infrastructure-heavy design problems.

How would you design the API surface for a client-side logging library, covering log levels, structured fields, and sampling?
Design the backend ingestion endpoint that receives log batches from clients.
How should the client library batch and flush logs without hurting foreground performance?
What happens when log volume is too high and the system can't keep up? How do you handle backpressure on both the client and server side?
How do you ensure that calling log() never blocks the caller, keeping it truly fire-and-forget?
Walk through the cross-cutting concerns for this system: schema evolution, PII scrubbing, authentication, offline behavior, and observability.

“Started with the obvious stuff, severity levels, key-value structured fields, and a trace ID.”

View Post

Snowflake·Software Engineer·Technical Phone Screen

Apr 2026

Snowflake technical phone screen with a grid simulation problem. Pretty niche question, not your typical BFS/DFS setup, and the edge cases are where it gets you.

You're given an n x m grid where a walker moves right one cell at a time, wrapping to the next row when it hits the last column. Some cells are obstacles, some are teleporters. Return the number of steps to reach the bottom-right cell, or -1 if an obstacle is hit, or -2 if the walker enters an infinite loop.

“The movement rule threw me off at first because it's not standard 2D traversal.”

View Post

Snowflake·Software Engineer·Onsite - System Design / Architecture

May 2026

System design round at Snowflake for a software engineering role. The whole session was built around designing an audit logs service for a multi-tenant SaaS platform, which sounds contained until you realize how many dimensions they want you to cover simultaneously.

Design the write path for an audit log service: how do producing services submit events, and how do you ensure durability without coupling producers to the storage layer?
How would you design the read path to keep p99 query latency low for per-account audit log queries with time range filters and pagination?
A single large tenant generates far more event volume than all others combined and is causing a hot partition. How do you rebalance without downtime or breaking time-ordered queries for that tenant?
How would you make the audit log tamper-evident to satisfy compliance requirements that events are provably unmodified since they were written?
Product wants to add free-text search and aggregations like 'all failed logins by IP this week' to the audit data. What changes to your storage or indexing strategy, and would you introduce a separate system?
A downstream consumer was offline for six hours during a high-volume period. How does your ingest design handle the catch-up without losing events or overwhelming storage?
What clarifying questions would you ask before designing this system, and how do the answers change your approach?

“I started with a Kafka-style ingest layer and the interviewer seemed satisfied with that direction pretty quickly.”

View Post

Snowflake·Software Engineer·Technical Phone Screen

May 2026

Snowflake SWE interview with a tree traversal problem that looked straightforward on the surface but had enough nuance to trip you up if you weren't careful about boundary conditions.

Given a full, complete, and balanced binary tree, return all boundary nodes in counter-clockwise order: root first, then the left boundary top-down, then leaves left to right, then the right boundary bottom-up. No duplicates.

“Variant of a known LC problem but the full+complete+balanced constraint changes things a bit.”

View Post

Snowflake·Software Engineer·Technical Phone Screen

May 2026

Snowflake software engineer screen, pretty much one meaty coding question that took up most of the time. The Unicode follow-up at the end was the part I didn't see coming.

Given a string, find the length of the longest substring with no repeated characters. Walk through an O(n) solution using a sliding window and hash map or set, and discuss the time and space complexity. How would you handle Unicode characters correctly?

“Classic question but the Unicode piece is where I fumbled.”

View Post

Snowflake·Software Engineer·Technical Phone Screen

May 2026

Snowflake SWE interview had me implement a full JSON parser from scratch, no library calls allowed. Pretty involved problem and the edge cases are where it gets brutal.

Implement a JSON parser from scratch that takes a raw string and returns the appropriate in-memory data structures, without using any built-in JSON parsing library.

“This one is bigger than it looks.”

View Post

Snowflake·Data Scientist·Technical Phone Screen

May 2026

Snowflake Data Scientist interview with a pretty deep algorithms question that felt more like a software engineering screen than anything data-related. One question, lots of moving parts, and they wanted both a recursive and iterative solution plus a full explanation of design choices.

Implement both a recursive and an iterative DFS on a directed graph that returns a topological order if the graph is a DAG, or detects and returns a cycle if not. Must also track discovery and finish timestamps for every node, handle disconnected graphs, self-loops, and parallel edges, all in O(N+M) time and space. Output as a dict with keys for is_dag, topo_order, cycle, discover, and finish.

“This was a lot.”

View Post

Snowflake·Software Engineer·Technical Phone Screen

May 2026

Two independent coding problems for a Snowflake software engineer round. Nothing too wild conceptually, but the second one had enough moving parts to slow me down.

Given a 1D array where 0 is empty space, 1 is a person, and 2 is a cake, return the minimum absolute distance between any person and any cake. Return -1 if either type is missing.
Given a start page and a target page in a wiki-like graph, return the minimum number of clicks to navigate from start to target by following outgoing links. You also need to implement a getLinks function that simulates retrieving links from a page.

“Pretty approachable once you just think of it as a two-pointer or brute-force scan.”

View Post