Google Professional Machine Learning Engineer
Get started today
Ultimate access to all questions.
Your team is developing a global banking application aimed at serving millions of users worldwide. The application features a sophisticated forecasting model designed to predict individual account balances three days in advance. These predictions will be used to trigger personalized notifications for users whose account balances are forecasted to fall below $25. Given the scale of the application and the need for real-time, efficient, and scalable notification delivery, what is the most effective method to serve these predictions? Consider factors such as scalability, cost-effectiveness, and the ability to deliver personalized notifications in real-time. Choose the best option from the following:
Your team is developing a global banking application aimed at serving millions of users worldwide. The application features a sophisticated forecasting model designed to predict individual account balances three days in advance. These predictions will be used to trigger personalized notifications for users whose account balances are forecasted to fall below $25. Given the scale of the application and the need for real-time, efficient, and scalable notification delivery, what is the most effective method to serve these predictions? Consider factors such as scalability, cost-effectiveness, and the ability to deliver personalized notifications in real-time. Choose the best option from the following:
Explanation:
The most efficient method to deliver these predictions is option D, leveraging Firebase Cloud Messaging (FCM) for its scalability, real-time notification capabilities, and efficient user management. FCM is specifically designed to handle high volumes of notifications, making it ideal for applications with millions of users. It ensures timely and personalized alerts by targeting specific users based on predictions from your model. Option E presents a viable alternative by combining FCM with Pub/Sub for enhanced scalability and real-time processing, but it introduces additional complexity. Options A, B, and C are less efficient due to either lack of personalization (A), scalability challenges (B), or unnecessary complexity and cost (C).