Professional-Machine-Learning-Engineer Leak Questions

Page: 21 / 21

Question 84

You need to develop a custom TensorRow model that will be used for online predictions. The training data is stored in BigQuery. You need to apply instance-level data transformations to the data for model training and serving. You want to use the same preprocessing routine during model training and serving. How should you configure the preprocessing routine?

Options:

Create a BigQuery script to preprocess the data, and write the result to another BigQuery table.

Create a pipeline in Vertex Al Pipelines to read the data from BigQuery and preprocess it using a custom preprocessing component.

Create a preprocessing function that reads and transforms the data from BigQuery Create a Vertex Al custom prediction routine that calls the preprocessing function at serving time.

Create an Apache Beam pipeline to read the data from BigQuery and preprocess it by using TensorFlow Transform and Dataflow.

Question 85

Your organization’s marketing team is building a customer recommendation chatbot that uses a generative AI large language model (LLM) to provide personalized product suggestions in real time. The chatbot needs to access data from millions of customers, including purchase history, browsing behavior, and preferences. The data is stored in a Cloud SQL for PostgreSQL database. You need the chatbot response time to be less than 100ms. How should you design the system?

Options:

Use BigQuery ML to fine-tune the LLM with the data in the Cloud SQL for PostgreSQL database, and access the model from BigQuery.

Replicate the Cloud SQL for PostgreSQL database to AlloyDB. Configure the chatbot server to query AlloyDB.

Transform relevant customer data into vector embeddings and store them in Vertex AI Search for retrieval by the LLM.

Create a caching layer between the chatbot and the Cloud SQL for PostgreSQL database to store frequently accessed customer data. Configure the chatbot server to query the cache.