Senior ML Engineer interview at Google
The KV Cache Memory Trap
“With a KV cache, each decode step only computes Q, K, V for the newest token. So it’s a free speedup, right? What did you actually trade away?”
Agent Engineering #33
Nov 5, 2026
Google interview traps
37 traps set in interviews at Google. Each one: the interviewer's question, the answer most candidates give, and the mechanism that breaks it.
Roles: Senior AI Engineer (13), Senior ML Engineer (8), Staff ML Engineer (6), Machine Learning Engineer (3), ML Engineer (2).
From RAG Interview Questions, AI Agent Engineering Interview, Advanced NLP Interview, LLM System Design Interview, Machine Learning System Design Interview, Advanced NLP Interview Questions, Computer Vision Interview Questions, LLM Inference Interview Questions and Advanced Reinforcement Learning Interview Questions.
Each trap is set in an interview at Google. AI Interview Prep isn't affiliated with Google.
Senior AI Engineer interview at Google
The Agent Self-Correction Trap
“Your research agent produced a full analysis report for $10. A PhD analyst needed 16 hours and $1,000 for the same work. The agent made one mistake: a single mis-pasted number. Your teammate suggests running it 5 times (still 20× cheaper) or having it re-read its own draft and self-correct. Will either one catch the error? What would you build instead?”
Agent Engineering #31
Nov 4, 2026Senior AI Engineer interview at Google
The Sequential Refinement Trap
“You built a sequential refinement pipeline. Agent 1 drafts, Agent 2 improves it, Agent 3 improves that. Agents 5 through 10 don’t move quality at all. Why is a chain the worst topology for scaling agent collaboration, and what’s the first structural change you’d make?”
Agent Engineering #29
Nov 2, 2026Senior ML Engineer interview at Google
The Causal Mask Leak
“Your autoregressive model hits a suspiciously low training loss, but every generation is garbage. What’s the first thing you check in attention? And how can a −inf mask vs. a ‘large negative’ mask still give you NaNs or leaks under mixed precision?”
Advanced NLP #11
Oct 26, 2026Senior ML Engineer interview at Google
The ReLU RNN Trap
“To fix vanishing gradients, a junior engineer opens a PR that swaps tanh for ReLU in your RNN. Their argument: ReLU’s derivative isn’t squashed below 1. Do you approve it?”
Advanced NLP #9
Oct 23, 2026Senior AI Engineer interview at Google
The Temperature 0 Trap
“Your agent makes different tool calls on identical inputs. A teammate says “set temperature to 0 and we’re done”
Advanced NLP #7
Oct 21, 2026AI Engineer interview at Google
The Front-End Verification Trap
“Your coding agent fixes a front-end bug. Every unit test passes. The UI is still visibly broken. How should the agent verify its own fix, and do you actually need a multimodal model to do it?”
Agent Engineering #23
Oct 12, 2026Senior AI Engineer interview at Google
The Unified Diff Trap
“Your coding agent has exactly one tool: bash. It can grep, sed, patch, and run pytest. So why does it keep botching file edits, and what edit interface would you give it instead of unified diffs with line numbers?”
Agent Engineering #21
Oct 10, 2026Senior AI Engineer interview at Google
The Agent Loop Trap
“A fixed localize → patch pipeline, with no tool loop and no verification step, beats your full coding agent on SWE-bench. Does that mean agent loops aren’t worth it?”
Agent Engineering #20
Oct 9, 2026Senior AI Engineer interview at Google
The Skill Evaluation Trap
“Your team shipped a code-review skill for your coding agent six months ago. Nobody knows if it helps. How do you build a production feedback loop that measures its quality from real PRs and turns that evidence into concrete skill edits?”
Agent Engineering #8
Sep 27, 2026Senior AI Engineer interview at Google
The Benchmark Autonomy Trap
“A year ago you shipped a fixed LLM workflow: hard-coded vendor list, fixed steps, pick the cheapest. A newer model now runs the whole task as a free-form agent and beats it on your benchmark. Do you delete the workflow?”
Agent Engineering #6
Sep 25, 2026Senior AI Engineer interview at Google
The Meta-Tool Trap
“Your agent has 70 REST tools and needs 12 sequential calls to compare smartphone prices across 4 countries. Switching to code execution raises success rates and cuts turns. So why isn’t code execution the default for every agent?”
Agent Engineering #1
Sep 20, 2026Staff ML Engineer interview at Google
The Tokenizer Swap Trap
“Your tokenizer gets 2.5 bytes/token on English but 1.1 on Vietnamese and Thai. Your teammate wants to grow the vocab from 100k to 250k. Do you ship it?”
LLM System Design #51
Aug 26, 2026Staff AI Engineer interview at Google
The AST Sandbox Trap
“You shipped a code-executing agent behind an AST sandbox that blocks os, subprocess, and dunder access. Security signed off. Six weeks later you’re breached. What class of attack did your sandbox structurally fail to stop?”
LLM Inference #13
Aug 11, 2026Senior AI Engineer interview at Google
The Multimodal Perception Trap
“Your web agent uses set-of-marks, you screenshot the page, draw numbered boxes on every element, and let the VLM click by number. It works in your demo but in production the model keeps clicking box 41 when it meant box 14, and it ignores half the page. The VLM is state-of-the-art and multimodal. So why is it failing?”
LLM Inference #6
Aug 4, 2026Senior ML Engineer interview at Google
The GraphRAG Scaling Trap
“Your GraphRAG pipeline uses an LLM to extract entity-relationship triplets. It’s flawless in the demo. Now you’re ingesting 50K docs a week and your indexing bill is on fire. What did you trade away when you picked LLM extraction?”
RAG #21
Jul 25, 2026Staff ML Engineer interview at Google
The Feedback Loop Trap
“Your REALM-style retriever learns from its own retrieval signal, no human labels. What breaks after six months in production?”
RAG #20
Jul 24, 2026Senior ML Engineer interview at Google
The Concatenation Trap
“Your conversational RAG has five query rewrite strategies. Why not just run all five and concatenate the outputs?”
RAG #19
Jul 23, 2026Senior ML Engineer interview at Google
The Hop-Count Paradox
“Adaptive RAG routes queries by predicted hop count, but nobody labels this question needs 2 hops. The paper bootstraps silver labels by brute-forcing every hop count and keeping whichever one got the right answer. What’s the ceiling that creates?”
RAG #18
Jul 22, 2026Senior AI Engineer interview at Google
The Flat Index Trap
“Your RAG pipeline scores 90% on factual lookups but collapses on ‘summarize this 300-page report.’ Why, and don’t tell me it’s chunk size?”
RAG #13
Jul 17, 2026Staff ML Engineer interview at Google
The Raw Vector Trap
“You and a teammate both ship IVF+PQ (Inverted File and Product Quantization) with identical memory budgets. Their recall is noticeably higher than yours. You quantized the raw vectors. They quantized the residual. Why does that one change beat you?”
RAG #10
Jul 14, 2026Staff ML Engineer interview at Google
The IVF-PQ Compression Trick
“Your RAG system hits 99% recall on HNSW in the demo. Now it’s 200M vectors and your RAM bill just triggered a budget review. Do you keep HNSW? Why or why not?”
RAG #8
Jul 12, 2026Senior AI Engineer interview at Google
The Prestige Trap
“Your RAG system pulls from a SQL database, a private company corpus, and web search. The answers now contradict each other. How do you fix source selection?”
RAG #7
Jul 11, 2026Search Infrastructure Engineer interview at Google
The Leaderboard Trap
“We benchmarked dense retrieval against BM25 on our eval set. Dense won on every metric. Do we rip out sparse retrieval entirely?”
RAG #5
Jul 9, 2026Staff ML Engineer interview at Google
The Parametric Memory Trap
“We want to swap our BM25 + dense pipeline for generative retrieval. Our corpus gains 50,000 documents every Monday. Why might that be a dealbreaker?”
RAG #4
Jul 8, 2026Staff ML Engineer interview at Google
The Semantic Chunking Trap
“Your RAG benchmark shows semantic chunking beating fixed-size by 8 points. You ship it. Production retrieval quality doesn’t move an inch. What happened?”
RAG #2
Jul 6, 2026Senior ML Engineer interview at Google
The Averaging Trap
“You’ve got three retrievers, BM25, a dense embedding model, and a rerank pass, and their relevance scores live on completely different scales. How do you merge them into one ranked list?”
RAG #1
Jul 5, 2026Senior AI Engineer interview at Google
The Gumbel-Softmax Trap
“If the discrete token bottleneck is the main reason we need RLHF, why not just use the Gumbel-Softmax trick to make sampling differentiable and backpropagate end-to-end?”
Reinforcement Learning #18
Feb 13, 2026Senior Computer Vision Engineer interview at Google
The L1 vs L2 Geometry Trap
“We’re building a similarity search for a new dataset. If I arbitrarily rotate the feature space by 45 degrees, which distance metric falls apart: 𝐋1 𝐨𝐫 𝐋2? And what does that tell you about our feature engineering strategy?”
Computer Vision #4
Jan 5, 2026Senior Computer Vision Engineer interview at Google
The Redundant Data Trap
“We trained a high-capacity ResNet on 500k images, but it’s still overfitting. My Product Manager wants to spend $20k to label another 500k random images scraped from the same source. Do you approve the budget?”
Computer Vision #2
Jan 3, 2026Senior ML Engineer interview at Google
The Inter-Annotator Agreement Trap
“We’re building a toxicity detection dataset where only 1% of comments are actually toxic. We hired two annotators. Their inter-annotator agreement is 99%. Are we good to go?”
NLP #22
Dec 27, 2025Senior AI Engineer interview at Google
The Perplexity Tokenizer Trap
“We ran an eval on a fixed dataset. Llama3 achieved a perplexity of 2.1, while Gemma3 scored 2.4. Which model is the better probability estimator, and which one do we deploy?”
NLP #18
Dec 24, 2025Machine Learning Engineer interview at Google
The Database-as-Queue Trap
“We have 3 upstream microservices generating features. They write to a central 𝘗𝘰𝘴𝘵𝘨𝘳𝘦𝘴 𝘋𝘉. Your ML Service queries that DB to get the input vector for inference. How do we scale this to 50k requests per second?”
ML System Design #19
Dec 5, 2025Machine Learning Engineer interview at Google
The Counterintuitive Truth About Quantization and Robustness
“Our edge model is vulnerable to adversarial noise, but we have strict latency limits. Should we avoid quantization (keeping Float32) to preserve model stability?”
ML System Design #15
Dec 2, 2025ML System Design #11
Nov 29, 2025ML Engineer interview at Google
The Two Faces of Inference
“We need to serve our model for two different use cases: a low-latency chatbot that needs a fast 𝐓𝐢𝐦𝐞-𝐭𝐨-𝐅𝐢𝐫𝐬𝐭-𝐓𝐨𝐤𝐞𝐧 (𝐓𝐓𝐅𝐓), and a high-throughput batch summarization job. How do these two workloads stress the GPU differently, and what fundamental tradeoff are you managing?”
LLM System Design #14
Nov 11, 2025ML Engineer interview at Google
The Gradient Highway
“Your team is struggling with training instability and exploding gradients in a new 100B+ model. The original ‘Attention Is All You Need’ paper used post-norm with learning rate warm-up. Why is that a bad idea for deep models, and what’s the one simple architectural change that solves this?”
LLM System Design #4
Nov 5, 2025