Snap Interview Questions — Page 9

Snap·Software Engineer·Onsite - System Design / Architecture

May 2026

System design round at Snapchat covering a ride-hailing backend, which sounds like an Uber prep question but had enough depth in the follow-ups to keep me on my toes. The scope was wide: order lifecycle, driver matching, user profiles, and peak load handling all in one session.

Walk through the full lifecycle of a ride request in a ride-hailing system, from the moment a rider submits a request to trip completion.
How would you design the driver-side flow for accepting or rejecting an incoming order, including timeout and reassignment logic?
Design the user profile storage for both riders and drivers, covering payment info, preferences, and any driver-specific data.
User profile reads happen at very high QPS. How do you design the read path to handle that load reliably?
How would you handle peak-hour load spikes? Cover caching strategy, capacity scaling, request throttling, and surge pricing as a demand-side lever.
Briefly, how would you approach geo-based driver matching? What data structures or services come into play?

“I started with the happy path and that was fine, but I underestimated how much they wanted to dig into state transitions.”

View Post

Snap·Software Engineer·Technical Phone Screen

May 2026

Snapchat coding round, graph theory question that sounds easy until you realize they want you to know two separate algorithms cold. Not a vibe check, they actually wanted implementation details.

Given a directed graph representing task or course dependencies, return a valid topological ordering of all nodes. If a cycle exists, return an indication that no ordering is possible. Be prepared to walk through both a BFS approach using in-degree tracking and a DFS approach using post-order traversal.

“I started with Kahn's algorithm because the queue-based approach feels more intuitive to explain out loud.”

View Post

Snap·Machine Learning Engineer·Technical Phone Screen

May 2026

Snapchat ML engineer screen, one coding question the whole time. They wanted a specific heap-based solution and would push back if you went a different direction, so knowing your complexity tradeoffs cold really mattered here.

Given an array of 2D points and an integer k, return the k points closest to the origin. Walk through an efficient solution and explain your complexity reasoning.

“The question itself is straightforward but they weren't just looking for any working solution.”

View Post

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

May 2026

ML system design round at Snapchat for an MLE role, focused entirely on building a lens recommendation system for a social camera app. Pretty deep dive, they pushed hard on the ML platform side which I wasn't fully prepared for.

Design a recommendation system for augmented-reality lenses in a social camera application, covering product goals, candidate generation, ranking, training data, and online serving.
How would you handle cold start for both new lenses and new users?
How would you use weighted logistic regression when different engagement actions like opens, creations, shares, and saves have different levels of value?
Describe what a feature store schema would look like for this system.
How would you correct for position bias in the training data?
How would you prevent the system from overexposing only the most popular lenses?
What should a machine learning platform include to support reproducibility and consistent feature usage between offline training and online serving?
What clarifying questions would you ask before starting the design, and what assumptions would you make if answers weren't available?

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

View Post

Snap·Product Manager·Onsite - Product Sense / Strategy

May 2026

Snap PM interview with a single product strategy question about TikTok monetization. Pretty open-ended, not much else to go on.

How would you approach monetization for TikTok?

“Interesting that Snap is asking about a competitor's business model.”

View Post

Snap·Software Engineer·Technical Phone Screen

May 2026

Snapchat SWE interview that went deep on networking fundamentals across three areas: real-time transport options, the OSI model, and TCP vs QUIC. The whole thing felt like a graduate networking course compressed into one session, and the follow-up questions were genuinely harder than the main ones.

Compare WebSocket, Server-Sent Events, and long polling for real-time web features. Cover directionality, how each connection is established, reconnection behavior, proxy and load balancer implications, and when you'd pick each one.
Walk through all seven OSI layers, what each one is responsible for, and give a real protocol or technology example per layer. Then explain how those seven map onto the four-layer TCP/IP model and why engineers use TCP/IP in practice.
Compare TCP and QUIC across handshake latency, TLS integration, multiplexing, head-of-line blocking, and connection migration. Include reasons you might NOT want to use QUIC.
If you have a fleet of stateless app servers and a user's WebSocket connection lives on one specific server, how do you push a message to that user from any server in the fleet?
QUIC supports 0-RTT connection resumption. What security risk does that introduce and how is it addressed?
A corporate firewall blocks all UDP traffic. Which of WebSocket, SSE, long polling, and QUIC still work, and how does the system degrade gracefully?

“I started with directionality which was the right call, but I fumbled the SSE establishment story.”

View Post

Snap·Machine Learning Engineer·Technical Phone Screen

May 2026

Snapchat ML Engineer interview that was basically a deep dive on Transformer internals. One long technical question, no fluff, they wanted you to actually know the architecture not just name-drop it.

Walk through the Transformer architecture from start to finish, covering input embeddings, positional encoding, multi-head self-attention with Q/K/V and masking, feed-forward layers, residual connections, LayerNorm, encoder vs decoder (including cross-attention), and how the final output projection produces predictions. Also explain why self-attention outperforms RNNs and what the O(n^2 * d) complexity means in practice.

“This is a lot to hold in your head at once.”

View Post

Snap·Software Engineer·Onsite - System Design / Architecture

May 2026

System design round at Snapchat, one big question that basically ate the whole hour. The scope was broad enough that I kept second-guessing how deep to go on each piece.

Design a ride-hailing service like Uber, covering the full stack: rider requests, driver matching, real-time tracking, the ride lifecycle, fare calculation, and payment.

“I started with the functional requirements which felt safe, but I spent way too long on the geo-indexing part (geohash vs quadtree vs S2) and never really got to payment consistency in any satisfying depth.”

View Post

Snap·Data Scientist·Technical Phone Screen

May 2026

Snapchat data scientist technical screen, pretty much a logistic regression gauntlet from start to finish. They went way deeper on the math than I expected, like full gradient and Hessian derivations, not just 'explain logistic regression at a high level.' Walked out unsure if I'd nailed it or completely embarrassed myself.

Write out the sigmoid function and the Bernoulli log-likelihood for binary logistic regression. Then derive the gradient and Hessian with respect to the coefficients under L2 regularization, and explain why the objective is convex.
With a single-feature model where the intercept is -1.2 and the coefficient is 0.8, compute the predicted probability for x = 2.0. Also report the odds and the odds ratio for a one-unit increase in x.
If the positive class occurs only 2% of the time and a false negative is ten times more costly than a false positive, what is the Bayes-optimal decision threshold? How would you calibrate model probabilities, and what are the tradeoffs between Platt scaling and isotonic regression?
What are the numerically stable forms for log(sigmoid(z)) and log(1 minus sigmoid(z)), and why do the naive expressions cause problems?
How does severe class imbalance affect maximum likelihood estimates in logistic regression, and what regularization or reweighting strategies would you apply? Back it up analytically.

“This is where I started sweating.”

View Post

Snap·Product Manager·Onsite - Product Sense / Strategy

May 2026

Snap PM interview with a product design question that sounds chill until you realize how much there is to unpack. Single question, open-ended, and the kind of thing where you can go in a hundred directions and still feel like you missed something.

How would you design a social media app specifically built for introverts?

“I started by questioning the premise a little, like what does 'for introverts' actually mean from a product standpoint.”

View Post