2023–2026 Google New Grad LLD / System Design: 10 Interview Questions

The verified Low-Level Design and System Design questions from 196 firsthand Google New Grad and Early Career interview experiences.

Coverage: Reports published from March 2023 through September 2026. The strict 196-experience corpus contains two LLD questions and eight System Design questions. Each question below has one verified sighting in this corpus.

Low-Level Design

01: Student room preference allocation

Design an allocation system that assigns students to rooms while handling room constraints and student preferences. The source preserved the design goal but not every final production rule.

Practice: Design Student Room Preference Allocation

02: Path-value file system

Design and implement a filesystem-like API that creates paths, stores values, and retrieves them. The interview focused on the object model, state transitions, invalid paths, and edge cases.

Practice: Design and Implement a Path-Value File System

System Design

01: Embeddable streaming chat agent

Design a chat-agent interface that can be embedded in another product. Reported discussion covered streamed responses, widget boundaries, chat history, caching, virtualization, and transport choices such as SSE or gRPC.

Practice: Embeddable Streaming Chat Agent Interface

02: Massive geospatial data visualization

Design a system that visualizes more than 100 million geospatial points, using a heat map at broad zoom levels and individual points when the user zooms in. Rendering, resolution, caching, and fallback behavior were part of the discussion.

Practice: Massive Geospatial Data Visualization

03: Distributed streaming skill matching

Design a distributed platform that ingests changing worker skills and job requirements, then produces suitable matches in real time.

Practice: Design Distributed Streaming Skill Matching

04: Real-time anomaly detection

Design a service that consumes a live event stream, detects anomalies, and produces timely results while handling scale, model evaluation, and operational failures.

Practice: Design a Real-Time Anomaly Detection Service

05: Harmful-content classification

Design an ML-backed service for detecting or classifying harmful content. The interview moved through data preparation, model choice, evaluation, serving, and feedback.

Practice: Design a Harmful-Content Classification Service

06: Online four-candidate ranker

At each interaction, choose one result from four candidates, observe user feedback, and improve future choices. Design the data, learning loop, evaluation, and serving path.

Practice: Design an Online Four-Candidate Ranker

07: Near-duplicate image detection

Design a service that detects exact or near-duplicate images at scale, including representation, candidate retrieval, similarity checks, thresholds, and evaluation.

Practice: Design Near-Duplicate Image Detection

08: Personalized video-rating prediction

Design a system that predicts how a user will rate a video and improves from later feedback. Cover feature and label collection, training, online serving, evaluation, and model updates.

Practice: Design Personalized Video-Rating Prediction

These are all ten verified LLD and System Design questions from the strict 196-experience corpus.