Google Technical Interview Questions

Surbhi G. - "In order to train a machine learning model for license plate detection, which is essentially image recognition, I would first need a large labeled dataset of images featuring license plates. From hereon, I would use a model architecture suitable for image-related tasks, like CNNs (Convolutional Neural Networks). When it comes to deploying the model, I would first optimize it, aiming for the best balance between speed and accuracy. Depending on the specifications, the model could be deployed eith"See full answer

AnonymousAxolotl - "Depends on what the ML model. If it's computationally heavy server side is best. If it's not, client side could be feasible, think of Apple Photos doing object recognition of photos on device, in this instance, the photos are on the device a long time and users are fine with waiting. So the use case important to consider. However broadly speaking, server side is most likely the best solution especially for this many users. Easier architecture to implement Example - client side you take"See full answer

Parth P. - "Establish more servers on regional bases or locally and move network traffic to these servers instead of keeping on one server."See full answer

Praful B. - "Set clear goals- What do you want to achieve by learning this new technology? Are you trying to build a specific project, or are you just interested in learning more about it? Having clear goals will help you stay motivated and focused. Find a good learning resource There are many different ways to learn a new technology. You can take online courses, read books, or watch tutorials. Find a resource that fits your learning style and budget. Start with the basics- Don't try to"See full answer

Adriel W. - "A DOS attack is meant to shut down a machine or network, making it inaccessible to its intended users. Some mitigations would be knowing what normal and abnormal traffic is. Deploy Firewalls for sophisticated application attacks."See full answer

Arvind R. - "Goal: Increase adoption of HPC capabilities offered by Google within enterprise customer context Vision: Accelerate high computing needs of enterprise, driven by innovation and definitive business outcomes Does this align with Google's vision - YES Trend Analysis: Technology: a. Introduce Intraday liquidity calculations b. Improve fraud detection using much large volume of data c. Faster value added risk calculations. d. Graph analysis for entity linking for large volume of customer a"See full answer

Kk - "Video is collection of images and we can call them as moving pictures. Let us break down the important components of the video in comparison to photos and explain how they impact : Frame rate per second : It is at which a total number of frames are captured in a video while it was shot and played then later Resolution : What resolution video was shot and played on devices ( Both can be different) Let us talk a bit deep dive about Frame rate per second. Often , a video co"See full answer

Abdurhman M. - "To determine what predictions to show, our systems look for common queries that match what someone starts to enter into the search box but also consider:"See full answer

Keval B. - "I would evaluate by comparing the results of old and new ranking algorithms, based on number of false positives."See full answer

Surbhi G. - "Understand the problem:Users in areas with poor or no data coverage can't fully utilize their smartphones. This affects communication, access to information, and use of various apps. Potential solutions: a) Offline capabilities: Enhance offline modes for popular apps (maps, messaging, content consumption). Implement better caching mechanisms to store frequently accessed data. Develop a system for queuing actions to be executed once coverage is restored. b) Improved connectivity: -"See full answer

Joohi M. - "Drew the following framework - Maturity of the tech (if it is too mature then consider buying), Competitors (how many, when are they launching this new tech etc.), Market Share of competitors, Financial considerations (build vs buy NVP), Collaborators for this tech, Economic Climate (Anti-trust, crisis etc.)"See full answer

Shubham S. - "Good afternoon everyone my name is shubham shamra, i'm from aligarh i've completed my graduation in electricak engineering from Gurukulkangri vishwavidyalaya haridwar. I having 6.5 year+ experience in project execution, operation and vendor and client management. Currently i'm working at vision india services pvt. ltd. here i'm managing smart metering project my role involves - leading team, coordinate with all DISCOM project manager and stake holder, billing and end to end operations. Befor"See full answer

Anonymous Muskox - "Metrics which Youtube Consider before building a recommender system Number of likes on a video by user The watch time of a video by the user The video disklied by the user The video share by a user The video skipped or churn with 20-30 seconds. Depending on this Youtube build a recommender system. The video suggestion feature in youtube works based on the recommender system. It may use a hybrid of batch prediction and online prediction. So depending on the above metrics, the youtube p"See full answer