Google MLE Interview Questions (2026)

Questions reported from Google machine learning engineer interviews. This bank tracks 46 questions asked at Google (16 with free model answers). Practice them live with DeepOffer's AI voice interviewer.

← All 465 questions

Practice this question with an AI interviewer

Get asked follow-ups live, then receive a scored report — like a real MLE interview loop.

Start AI mock interview

#1 Explain a Single K-Means Update Step by Step ★
#2 How Can Softmax Be Calculated Without Numerical Overflow? ★
#8 How Should You Explain the Trade-Off Between Model Bias and Variance? ★
#18 Architect the Retrieval and Ranking Stack for Search ★
#19 Architect a Low-Latency Model for Predicting Ad Click Probability ★
#23 Describe a Time You Resolved a Professional Disagreement ★
#35 Given Item Counts, How Would You Return the K Most Frequent Values? ★
#36 How Would You Build an O(1) Least-Recently-Used Cache? ★
#37 How Would You Code Stable Softmax, Cross-Entropy, and Their Gradient in NumPy? ★
#40 Architect a Production Service for Autoregressive LLM Generation ★
#53 Compare greedy, beam, top-k, top-p, and temperature decoding, and identify when each one fails. ★
#76 A GPU/TPU rated at 500 TFLOPS sustains only ~50 TFLOPS on a large-model kernel — what chip-level factors could explain the 10x drop?
#78 Compare fine-tuning T5, RoBERTa, and Llama 2 7B for multi-turn conversational chatbots and argue which is most likely to perform best.
#103 Define the bias-variance trade-off and discuss how the two relate.
#104 State the assumptions of linear regression.
#105 Distinguish regularization from validation and explain when each is the right tool.
#106 Derive the gradient of cross-entropy loss with softmax inputs and explain why the two are fused numerically.
#107 Explain the SVD and give two places it appears in modern deep learning.
#108 On average, how many fair coin flips until two heads in a row? Walk through the reasoning.
#109 Decide when Q-learning beats policy gradients and vice versa.
#110 With only limited access to feature weights, how would you explain a loan rejection from a binary classifier?
#116 Explain the bias-variance trade-off, cross-validation, and the curse of dimensionality together.
#118 Explain when you would prioritize precision, recall, or F1 over the others.
#131 Explain precision and recall, and how to avoid underfitting and overfitting.
#171 Compare tensor, pipeline, data, sequence, and expert parallelism and explain when to combine them. ★
#211 Diagnose and fix a pretraining loss that suddenly diverges at step 300k of a long run.
#212 Design the training setup for a 70B-parameter model that does not fit on one accelerator.
#235 Defend RAG over fine-tuning for an enterprise assistant, and say where RAG breaks.
#271 Design an ML system that predicts which links to show for a search query, given user_id, timestamp, query, and clicked links.
#296 Implement a correct multi-head self-attention forward pass with projections, reshaping, and masking. ★
#300 Implement top-k, top-p, and temperature sampling over a logits vector. ★
#305 Implement k-means clustering on a shared doc. ★
#325 Find the top-K most frequent items in a high-volume event stream with bounded memory.
#326 Return the k most frequent event IDs from an unbounded stream at any point, with bounded memory. ★
#365 Describe a time you disagreed with a researcher or tech lead about priorities, and what happened.
#392 Describe your favorite product and explain why.
#417 Design an image-generation service (Midjourney/DALL-E style): queue management, GPU scheduling, and content-safety filtering at scale
#418 Design an autonomous travel-booking agent: orchestrator, tool gateway, credential management, rollback, and risk-classified actions
#419 Design a hospital voice assistant: real-time STT+LLM+TTS pipeline, HIPAA compliance, and medical latency budgets
#427 Design an ML experiment-tracking platform: run logging, metric comparison, and artifact management at org scale
#430 Design automatic face and license-plate blurring for street-level imagery at planetary scale
#435 Design a hallucination-free banking chatbot: domain-constrained generation, citation verification, and regulatory compliance
#438 Implement TF-IDF vectorization from scratch with NumPy
#439 Implement beam search decoding over next-token logits from scratch
#440 Implement the batch normalization forward and backward pass in NumPy
#441 Implement precision, recall, and F1 from raw predicted labels in NumPy, without sklearn