Verily interview questions and how to prepare

1. Imagine Verily is developing a new AI model to predict the onset of diabetic retinopathy from retinal scans. Describe the system architecture for ingesting, processing, and serving predictions for millions of patient scans globally, considering latency, scalability, and data privacy. What are the key trade-offs you'd consider in choosing between a microservices and a monolithic architecture for this system?

   Software EngineerSystem DesignTests: Evaluates the candidate's ability to design scalable, distributed systems for complex AI workloads, considering trade-offs and real-world constraints.

2. A critical service in Verily's cloud infrastructure, responsible for anonymizing patient data before it's used for research, is experiencing intermittent high latency. How would you approach debugging this issue? Walk me through your thought process, the tools you'd use, and what specific metrics you'd monitor.

   Software EngineerDebuggingTests: Assesses problem-solving skills, systematic debugging approach, and familiarity with common cloud infrastructure issues.

3. Verily is building a platform for continuous glucose monitoring (CGM) data analysis. You're tasked with implementing a real-time anomaly detection algorithm for glucose spikes. Write a Python function that takes a time series of glucose readings and identifies significant deviations from the expected trend, using a simple moving average and standard deviation approach. Discuss how you would handle edge cases like missing data points or sensor drift.

   Software EngineerCoding and ReasoningTests: Tests coding proficiency, algorithm implementation, and ability to handle real-world data imperfections.

4. Verily's Study Watch collects various physiological data. Design a system to efficiently store and query time-series data from millions of Study Watch users for research purposes. Consider data modeling, indexing strategies, and potential performance bottlenecks. How would you handle data retention policies and GDPR compliance?

   Software EngineerSystem DesignTests: Probes understanding of large-scale data storage, time-series databases, and compliance considerations in healthcare data.

5. You're designing a new feature for Verily's Onduo platform that allows users to share their health data with their physicians. Describe the API design for this data sharing functionality, focusing on security, privacy, and versioning. What are the key considerations for handling different data formats and permissions?

   Software EngineerSystem DesignTests: Evaluates API design skills, understanding of security and privacy principles, and ability to design for interoperability.

6. Verily is developing a tool to help researchers identify potential drug targets by analyzing large genomic datasets. You need to implement a feature that performs k-means clustering on gene expression data. Write a Python snippet to perform k-means clustering using a library like scikit-learn, and then explain how you would scale this to handle terabytes of data. What are the potential challenges and solutions?

   Software EngineerCoding and ReasoningTests: Tests coding skills, familiarity with ML algorithms, and understanding of scaling data processing challenges.

92+ more Verily-style questions are in the free library, each practiceable live with adaptive follow-ups and an honest scorecard. Start free.