An enterprise analytics team needs to enrich several terabytes of unstructured customer feedback records already stored in BigQuery standard tables. The team wants to leverage a Gemini foundation model hosted on Vertex AI to perform sentiment summarization and structured entity extraction. The solution must satisfy the following criteria: Execute large-scale batch inference directly on data residing in the warehouse using standard SQL Avoid moving or exporting data to external intermediate storage Minimize operational overhead by eliminating the need to manage dedicated endpoint virtual machines or custom serving infrastructure Which architecture should the data engineering team implement?

Create a BigQuery ML remote model referencing the Vertex AI Gemini model via a Cloud resource connection, and run queries using the ML.GENERATE_TEXT function directly on the analytical tables.: ### What BigQuery ML Remote Models with Vertex AI Are **BigQuery ML remote models** enable BigQuery to orchestrate machine learning inference using models deployed to or hosted in **Vertex AI**, including foundation models like **Gemini**. By establishing a secure **Cloud resource connection**, BigQuery delegates the model execution to Vertex AI while managing query execution and output ingestion automatically. ### How It Meets the Requirements - **Direct in-warehouse inference**: SQL queries invoke the **`ML.GENERATE_TEXT`** function directly against standard BigQuery tables, passing table columns as prompts or structured inputs. - **No intermediate data movement**: Data does not need to be exported to Cloud Storage or routed through staging platforms; all transformations and evaluations happen seamlessly inside the warehouse environment. - **Zero infrastructure management**: BigQuery ML abstracts the underlying API integration and scaling mechanics, eliminating the need to provision, size, or autoscale dedicated **Vertex AI prediction endpoints** or worker pools. ### Key Advantages and Capabilities - **Seamless SQL workflow**: Analysts and data engineers can execute generative AI enrichments using familiar **GoogleSQL** syntax. - **Serverless scaling**: BigQuery automatically batches and orchestrates calls to **Vertex AI**, returning results directly into destination tables. - **Enterprise governance**: Leverages **Identity and Access Management (IAM)** and dataset permissions without exposing raw API keys or managing custom client code. ### Why This Is Optimal Executing inference directly inside the data warehouse via BigQuery ML remote models provides the lowest **operational complexity** and the highest developer productivity for analytical batch workloads. Compared to extracting data or deploying inline pipeline predictors, it avoids unnecessary data duplication and infrastructure overhead.

Bigquery Ml Remote Models Vertex Ai Batch Inference | professional-cloud-data-engineer

Bigquery Ml Remote Models Vertex Ai Batch Inference

Worked example. The correct answer is already marked and every option is explained below, so there is nothing to select here. To answer questions yourself, start the free trial.

An enterprise analytics team needs to enrich several terabytes of unstructured customer feedback records already stored in BigQuery standard tables. The team wants to leverage a Gemini foundation model hosted on Vertex AI to perform sentiment summarization and structured entity extraction.

The solution must satisfy the following criteria:

Execute large-scale batch inference directly on data residing in the warehouse using standard SQL
Avoid moving or exporting data to external intermediate storage
Minimize operational overhead by eliminating the need to manage dedicated endpoint virtual machines or custom serving infrastructure

Which architecture should the data engineering team implement?

Deploy a custom Dataflow streaming pipeline that reads records, executes synchronous inline HTTP inference calls to an autoscaled Vertex AI online prediction endpoint, and writes the output back to BigQuery.

Export the BigQuery tables to Cloud Storage as JSON files, trigger an asynchronous Vertex AI Batch Prediction job with custom worker pools, and load the results back into BigQuery.

Create a BigQuery ML remote model referencing the Vertex AI Gemini model via a Cloud resource connection, and run queries using the ML.GENERATE_TEXT function directly on the analytical tables.

Train an XGBoost text classification model within BigQuery ML, register it in Vertex AI Model Registry, and deploy it to a Vertex AI endpoint to execute predictions.

Professional Cloud Data Engineer

Bigquery Ml Remote Models Vertex Ai Batch Inference

Related questions