← Back to Directory

Snap interview notes, page 6 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·Product Manager·Onsite - Product Sense / Strategy

Apr 2026
PM interview at Snap, one product sense question about improving an existing feature. Pretty focused session, no fluff.
  • How would you improve the Spotlight feature at Snapchat?

“I went straight into user pain points and started rattling off ideas before really anchoring on what Spotlight is even trying to do versus TikTok or Reels.”

View Post

Snap·Data Scientist·Technical Phone Screen

Apr 2026
Snapchat data scientist interview that leaned heavily on probability fundamentals. The Bayes' theorem question was the centerpiece and required a full worked example from scratch, not just a formula recitation.
  • Walk through a complete Bayes' theorem calculation using a concrete example. Define your prior and likelihoods, write out the full formula, then compute the posterior and give a final numeric answer.

“I went with the classic disease-test setup because it's easy to plug numbers into, but I fumbled the denominator for a bit.”

View Post

Snap·Software Engineer·Technical Phone Screen

Apr 2026
Snapchat SWE interview that leaned hard into graph/tree problems. The main challenge was a BFS/DFS hybrid question with a follow-up on squeezing space complexity down, which I did not see coming.
  • Solve a graph or tree traversal problem using a combination of BFS and DFS, then explain how you would optimize space from O(N) to O(1) by reusing the input structure or marking visited states in-place.

“I got the BFS/DFS part working fine but then they asked me to drop the visited set entirely and encode state directly into the input.”

View Post

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

May 2026
Snapchat ML engineer interview that went deep into multimodal retrieval systems. The whole session was basically one long design question with a lot of follow-ups, and they clearly wanted you to have strong opinions, not just recite concepts.
  • Design a short-video recommendation and retrieval system using ML, given that only 20% of the videos have any text description attached.
  • How do CLIP-style joint vision-text embeddings work, and why are they useful here?
  • Walk through the loss functions used in contrastive learning, specifically InfoNCE and NT-Xent, and explain why they're used in this context.
  • What are the main weaknesses of pure embedding-based retrieval, and how would you address them?
  • How would you combine sparse and dense retrieval, and where does a reranker fit in?
  • How do you handle popularity bias in a retrieval or recommendation system, and what techniques would you use to mitigate it?

“This is a meaty one.”

View Post

Snap·Machine Learning Engineer·Technical Phone Screen

May 2026
Snapchat ML Engineer interview that went deep on neural network fundamentals. Three main areas: loss functions, optimizers, and architectures, each with follow-up questions that pushed pretty far below the surface. Not a vibe check, they actually wanted to see if you understood the math.
  • Walk me through the main loss functions used in regression and classification. When would you pick one over another, and what does the gradient behavior look like for each?
  • How do contrastive and triplet losses work, and when would you reach for a ranking loss instead of a standard classification objective?
  • Compare SGD with momentum, RMSProp, and Adam. What are the practical tradeoffs in terms of convergence speed and how they affect generalization?
  • Explain learning rate schedules: warmup, cosine decay, and step decay. Why does warmup matter at the start of training?
  • Describe how self-attention works in a Transformer, including multi-head attention and positional encoding. Why do we need positional encoding at all?
  • When would you use a CNN versus an RNN versus a Transformer for a sequence modeling problem?

“This started easy enough, MSE versus MAE, then they asked about Huber and I fumbled a bit explaining the transition point between the two regimes.”

View Post

Snap·Software Engineer·Onsite - System Design / Architecture

May 2026
System design round at Snapchat, basically a full Uber clone from scratch. Dense question with a lot of moving parts and I definitely ran out of time before covering everything cleanly.
  • Design a ride-hailing service like Uber. Walk through the full system: how riders request rides, how drivers are matched in real time, ETA and live tracking, fare calculation, payments, and ride history.
  • How would you handle driver location updates at city scale, and what geo-indexing approach would you use for low-latency driver matching?
  • How do you handle surge pricing and peak-hour hot spots without creating bottlenecks in the system?
  • Walk through the high availability and partitioning strategy for a service operating across multiple cities simultaneously.

“I started with the functional requirements which felt like the right move, but I spent way too long on the rider-facing flows and barely got to the interesting stuff.”

View Post

Snap·Technical Program Manager·Onsite - System Design / Architecture

May 2026
Snapchat TPM interview focused heavily on ML platform reliability and the messier operational side of the role. Less about coding, more about whether you've actually dealt with a production fire and lived to tell about it. The follow-up questions were sharp and pushed past surface-level answers pretty fast.
  • If an ML service's SLA suddenly drops, how do you identify the root cause and what steps do you take to stabilize and improve reliability afterward?
  • How do you evaluate the ROI or cost savings of a reliability or infrastructure investment?
  • How do you hold cross-functional teams accountable for reliability commitments without turning it into a blame culture?
  • What do you do when your headline metrics look healthy but leadership is still dissatisfied with the service?
  • How would you prioritize capacity work versus model quality improvements when both are competing for the same team's bandwidth?
  • Your average SLA looks fine but enterprise customers are complaining. How do you respond?

“This was the core question and it sprawled.”

View Post

Snap·Software Engineer·Onsite - System Design / Architecture

May 2026
System design round at Snapchat, basically a deep dive into how Stories works under the hood. Covered a lot of ground fast and I don't think I handled the fanout section as well as I should have.
  • Design Snapchat Stories: users post short photo/video clips visible to friends for 24 hours, then they disappear. Walk through your full system design.
  • How would you handle media upload and delivery at scale for Stories?
  • How do you store story metadata with a 24-hour TTL, and what happens when stories expire?
  • For the friends' Stories feed, would you use a push or pull fanout model? Justify your choice.
  • How would you track which stories a user has already viewed, at scale?
  • How do you enforce privacy and visibility rules for Stories, for example close friends only vs all friends?

“I started with requirements and that part went okay.”

View Post

Snap·Software Engineer·Technical Phone Screen

May 2026
Snapchat software engineer interview that came down to a classic multi-list merge problem. Pretty algorithmic, no fluff, they wanted to see if you knew your heap mechanics cold.
  • Given N sorted lists, merge them all into a single sorted output. Walk through your approach and analyze the time and space complexity.

“Knew this one but still fumbled the explanation a bit at first.”

View Post

Snap·Data Scientist·Technical Phone Screen

May 2026
Snapchat data scientist round focused entirely on Bayesian statistics applied to a spam-detection system already running in production. Pretty deep technically, more than I expected for a single session. Left feeling like I held my own on the theory but stumbled a bit when the conversation shifted to live monitoring.
  • For a Beta-Binomial click-through model, explain the roles of the prior, likelihood, and posterior. Then derive how the posterior mean works as a smoothed estimate and walk through when you'd prefer it over MLE.
  • How do Bayesian credible intervals differ from frequentist confidence intervals, specifically in the context of calibrating a production model?
  • You detect concept drift in the production model. How would you update the prior or the model hierarchy to make it adapt faster?

“I actually enjoy this stuff so the derivation part went fine.”

View Post