← Back to Directory

Google interview notes, page 110 of 215

FAANG / Big Tech

Most of these Google write-ups come from Software Engineers (1,167), then Product Managers (686) and Data Scientists (125). The most reported rounds are the product sense round (667), the technical phone screen (660), the behavioral / leadership round (252) and the onsite coding round (214).

2146 interview notes · updated Oct 2026

Google·Machine Learning Engineer·Technical Phone Screen

May 2026
Coding round for an ML Engineer role at Google. The question looked like a standard Python exercise but had enough edge cases baked in that it turned into a pretty involved conversation about generator design and test coverage.
  • Write a Python generator that yields integers from a list, handling None values, empty input, duplicates, very large inputs, and non-integer items. You need to define and document what the generator does with invalid entries (skip, convert, or raise), then write comprehensive unit tests covering normal behavior, edge cases, iteration order, resource usage, and error handling.

“The generator part itself wasn't the hard bit.”

View Post

Google·Product Manager·Onsite - Product Sense / Strategy

May 2026
PM interview at Google with a classic estimation question. Pretty short and to the point, nothing wild to report.
  • How many grains of rice are in a 10kg bag?

“Classic fermi estimation.”

View Post

Google·Product Manager·Onsite - Product Sense / Strategy

May 2026
Google PM interview with a single estimation question about traffic and lost productivity. Pretty classic fermi problem but it has a few moving parts that can trip you up if you're not careful about your assumptions.
  • Estimate the total number of productivity hours lost per year due to traffic congestion.

“I started by trying to scope it geographically and immediately second-guessed myself on whether to do US-only or global.”

View Post

Google·Machine Learning Engineer·Technical Phone Screen

May 2026
Deep technical screen for an ML Engineer role at Google. The whole thing was basically one massive question about Transformers from first principles all the way through to molecular applications. Left feeling like I'd just taken an oral exam on a graduate ML course.
  • Walk through the Transformer architecture in detail, covering the encoder and decoder stack structure, self-attention, cross-attention, and position-wise feed-forward networks.
  • Derive the scaled dot-product attention equation and explain the shapes of the key, query, and value matrices. Then explain how multi-head attention extends this.
  • Compare sinusoidal and learned positional encodings. How do relative position schemes differ, and what does any of this mean for how the model understands sequence order?
  • What are the differences between encoder-only, decoder-only, and encoder-decoder Transformer models? How does masking work in autoregressive decoding?
  • The standard Transformer has O(n²) complexity in sequence length. What approaches exist for scaling to longer sequences, and what trade-offs do they introduce?
  • Discuss pre-LN versus post-LN placement of LayerNorm in Transformer blocks. How do residual connections, training stability, and initialization interact here?
  • How would you adapt a Transformer to work with molecular data like SMILES strings or molecular graphs? Cover tokenization, stereochemistry, data augmentation, and what training objectives make sense.

“I started with the encoder side and worked my way through, but I fumbled when they pushed on cross-attention specifically.”

View Post

Google·Software Engineer·Technical Phone Screen

May 2026
Google SWE coding round, basically a Number of Islands variant. Pretty standard graph traversal problem but I still managed to second-guess myself halfway through.
  • Given a 2D grid of '1's (land) and '0's (water), count the number of islands where an island is a group of horizontally or vertically connected land cells. Implement a solution in Python using DFS or BFS.

“Went with DFS.”

View Post

Google·Software Engineer·Onsite - Coding / Algorithms

May 2026
Interviewed at Google for a SWE role. The content shared was just a title, 'Degrees of Friendship', which sounds like a graph problem, probably BFS or shortest path between nodes in a social network. Not much else to go on.
  • Given a social network represented as a graph, find the degree of connection (shortest path) between two people.

“Classic BFS setup.”

View Post

Google·Software Engineer·Onsite - Coding / Algorithms

May 2026
Google coding round, pretty focused on data structure design. One question but it had a lot of layers and the follow-up about heaps came out of nowhere.
  • Design a Bookshelf class that supports adding a book, moving a book to a different position, and deleting a book. Handle edge cases like out-of-range indices, duplicate titles, and moving a book to its current position. Then discuss how you'd swap out the underlying structure for a priority heap if ordering by some priority became a requirement.

“I went straight to an array with splice and felt pretty good about it until they started poking at edge cases.”

View Post

Google·Product Manager·Onsite - Product Sense / Strategy

May 2026
Google PM interview covering the fundamentals: what skills and knowledge a PM actually needs, and how you'd go about understanding your users. Nothing too exotic but the questions are deceptively open-ended.
  • What are the core skills and types of knowledge a PM should have?
  • As a PM, what strategies would you use to understand your users?

“Broader than it sounds.”

View Post

Google·Software Engineer·Online Assessment (OA)

May 2026
Google SWE online assessment with a bit math/greedy problem. Nothing too wild but the constraint range (up to 10^18) means you can't brute force it, which is where most people probably trip up.
  • Given a positive integer n (up to 10^18), find the minimum number of operations to reduce it to 0. If n is even you divide by 2; if n is odd you can either add 1 or subtract 1.

“The small examples make it feel easy and then you see the constraint is 10^18 and suddenly brute force or recursion with memoization starts looking sketchy.”

View Post

Google·Product Manager·Onsite - Product Sense / Strategy

May 2026
Google PM interview that opens with a product likes/dislikes exercise and then drills pretty deep on one product you pick. The structure sounds casual at first but it escalates fast into metrics and vision territory.
  • Name three products you like and three you dislike, then pick one to discuss in depth.
  • Why do you like or dislike the product you chose?
  • What would you change about this product?
  • How would you prioritize the changes or features you just suggested?
  • How would you measure whether the feature you prioritized is actually successful?
  • If you had unlimited budget as the PM of this product, where would you take it over the long term?

“The setup feels like a warmup but it's really the foundation for everything that follows.”

View Post