Snap interview notes, page 22 of 22

Large Enterprises

Most of these Snap write-ups come from Software Engineers (117), then Machine Learning Engineers (65) and Data Scientists (18). The most reported rounds are the technical phone screen (149), the system design round (45), the behavioral / leadership round (11) and the product sense round (8).

216 interview notes · updated Jul 2026

Snap·Machine Learning Engineer·Technical Phone Screen

Jul 2026

Snapchat ML Engineer coding round, two problems back to back. Both were more classic CS than anything ML-specific, which surprised me a little.

Build an arithmetic expression evaluator that parses a string with non-negative integers, spaces, the four basic operators, and parentheses. Operator precedence applies and division truncates toward zero.
Implement a Union-Find solution to find all connected components in an undirected graph. Return each component as a sorted list of node IDs.

“This one took me a minute to even figure out where to start.”

View Post

Snap·Software Engineer·Technical Phone Screen

Jul 2026

Snapchat SWE interview with a classic dynamic programming problem. Pretty standard coding round, nothing too surprising, but the edge cases around zeros will trip you up if you're not careful.

Given a string of digits, count the total number of ways to decode it into letters using the mapping where '1' maps to 'A', '2' to 'B', and so on up to '26' mapping to 'Z'. Substrings like '06' or '30' are invalid.

“The base logic clicked pretty fast for me, single digit or two digit, build up a dp array.”

View Post

Snap·Technical Program Manager·Technical Phone Screen

Jul 2026

Interviewed for a TPM role on the ML Platform team at Snap. The session was heavily technical for a program management role, covering ML fundamentals and a full lifecycle design exercise. More depth than I expected going in.

Walk me through the key ML evaluation concepts a TPM should know: recall, ROC curve, AUC, p-value, hypothesis testing, and Type I versus Type II errors.
How would you explain AUC to a non-technical stakeholder?
Offline AUC improved significantly in your model evaluation, but after launching the experiment, retention is dropping. What do you do?
Design the end-to-end ML lifecycle for a content recommendation service at Snap, covering system components, stakeholders, launch metrics, and post-launch monitoring.
Who are the key stakeholders for an ML-powered recommendation launch and how do you manage competing priorities across them?
How would you diagnose model drift after a recommendation model has been in production for several weeks?
How do you determine whether an experiment is underpowered before concluding it failed?

“I knew most of this but stumbled on explaining AUC in plain language.”

View Post

Snap·Machine Learning Engineer·Technical Phone Screen

Jul 2026

Senior ML loop at Snapchat focused almost entirely on one deep-dive recommendation project. No LeetCode, just a long technical conversation about modeling decisions and tradeoffs. Left feeling like I'd either nailed it or completely exposed myself, still not sure which.

Walk me through the modeling approach you chose for a past recommendation or prediction project and explain why you made that choice.
Why didn't you go with a more complex model for that project?
How do you think about balancing precision, recall, latency, and infrastructure cost when making modeling decisions?
What would you do if your offline evaluation metrics look good but the online business metrics are moving in the wrong direction?
How would you go about tuning the decision threshold for a binary classifier in production?
Compare Factorization Machines with embedding-based deep neural networks for a recommendation or ranking task. When would you pick one over the other?

“I anchored on a two-tower retrieval model I'd built before and talked through why it made sense for the scale we were at.”

View Post

Snap·Software Engineer·Technical Phone Screen

Jul 2026

Snapchat software engineer interview with two fairly distinct coding problems. One was a geometry/graph problem and the other was a tree-building exercise with a follow-up on mutations. Nothing too wild but the DOM question had more depth than I expected.

You're given a set of 2D circles representing stones in a river. Determine whether the stones collectively block the full width of the river. You can define your own input format.
Given a sequence of HTML-like tokens (for example: open paragraph, raw text 'ABC', close paragraph), parse them and build a DOM tree, then print it. Follow-up: add support for inserting and deleting nodes.

“The 'define your own input format' part threw me a little, felt like a trick but it's probably just them being flexible.”

View Post

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

Jul 2026

Snapchat MLE system design round, heavy on geo and ML pipeline stuff. The question was basically one giant beast that kept spawning sub-problems every time I thought I was done. Walked away feeling okay about the spatial indexing parts but less sure about how I handled the ANN vs. learned ranking discussion.

Design a real-time system that surfaces the top-K points of interest within a user's current map viewport as they pan and zoom, with p95 end-to-end latency under 150ms.
When would you use approximate nearest neighbor search for candidate retrieval versus a learned ranking model for re-ordering results, and how do you decide which to use?
How would you handle load shedding, deduplication across tile boundaries, and pagination while the user is actively panning?
How would you evaluate this system both offline and online, and what metrics would you track?
How do you handle cold start for new POIs or new users in this system?

“This one sprawled in every direction.”

View Post