PMLE practice questions 331–340

1. Question 331 · 364 in bank

   You work at a manufacturing company. You are building an assistant to help engineers write scripts for a proprietary, legacy machine control language called “MechScript.” MechScript is highly specialized, uses unique syntax not found in public datasets, and is strictly internal to your company. You built the base Gemini model using Agent Platform RAG Engine and grounded the model with the MechScript reference manual. The model correctly answers questions about the language, but it consistently fails to generate syntactically correct code snippets, often hallucinating Python-like syntax instead. What should you do?

   • AIncrease the temperature parameter in the Gemini API call.
   • BUse the Agent Platform supervised fine-tuning (SFT) service with a dataset of valid MechScript examples.
   • CIncrease the number of retrieved chunks in the RAG Engine call.
   • DCopy the MechScript manual into the system prompt using Gemini’s context window.

   Show answer

   Answer: B

   Next question

2. Question 332 · 364 in bank

   Your company’s data science team has developed a complex feature engineering workflow using Python with pandas that runs on a single Agent Platform Workbench instance. Due to a recent increase in dataset size, the script is failing with out of memory (OOM) errors. You need to scale this workload to run across multiple nodes to handle the data volume. The data science team wants to use Python-native APIs and minimize the effort required to rewrite the code or manage infrastructure. What should you do?

   • AContainerize the script. Create a pipeline on Agent Platform Pipelines using the Kubeflow Pipelines (KFP) SDK that executes the container on a machine type with higher memory.
   • BProvision a Google Kubernetes Engine (GKE) cluster. Deploy the dask-kubernetes operator. Update the script to use Dask for distributed dataframe operations that are managed by the team.
   • CUse the Agent Platform SDK to create a Ray on Agent Platform cluster. Update the script to use Ray Data libraries for distributed processing, and submit the job to the cluster.
   • DUpdate the script to use Apache Beam. Run the job on Dataflow.

   Show answer

   Answer: C

   Next question

3. Question 333 · 364 in bank

   You work for a large online retailer. You are productionizing a demand forecasting model designed to predict future sales for inventory management. You need to retrain the model with fresh data every 30 days. The inventory planning team requires daily batch predictions for replenishment orders. You need to design a solution that generates daily predictions while minimizing resource costs. Which pipeline architecture should you implement?

   • ASchedule a pipeline in Agent Platform Pipelines that generates batch predictions daily. Include training as a conditional step that runs if it is day 30 of the cycle.
   • BImplement two pipelines in Agent Platform Pipelines: a continuous training pipeline that runs every 30 days, and a daily batch prediction pipeline that runs the batch prediction job.
   • CImplement a pipeline in Agent Platform Pipelines for training that runs every 30 days to update a versioned model on an Agent Platform endpoint. Configure a Cloud Run function to send the daily batch dataset to the endpoint for predictions.
   • DImplement a pipeline in Agent Platform Pipelines with sequential steps for data preparation, model training, and batch prediction. Schedule the pipeline to run daily.

   Show answer

   Answer: B

   Next question

4. Question 334 · 364 in bank

   You are developing a Gemini-based onboarding chatbot to answer users’ questions based on five PDF onboarding documents in Cloud Storage. The documents are updated weekly or monthly as needed. You expect that the application will receive high-volume, repeated user queries against this documentation context. You need to use the most appropriate and cost-effective solution for managing the context and ensuring that the chatbot has up-to-date information. What should you do?

   • AFine-tune a Gemini model using the text from the five PDF files as a training dataset.
   • BSend requests with the five PDF files as content pointing to the URI file. Use implicit context caching.
   • CCreate an explicit context cache using the five PDF files, and regenerate the cache when the source files change.
   • DCreate an explicit context cache using the five PDF files. Set a time-to-live (TTL) of seven days, and regenerate the cache every seven days.

   Show answer

   Answer: C

   Next question

5. Question 335 · 364 in bank

   You work for a logistics company that receives thousands of PDF invoices each day from various freight carriers. The invoices are scanned and saved as images in a Cloud Storage bucket. Each invoice contains important information such as invoice number, invoice date, supplier name, and invoice amount. However, the layout and format vary significantly from carrier to carrier. You need to build a pipeline that extracts the fields from all incoming invoices and loads the structured data into a BigQuery table for analysis. You want the solution to be scalable while also minimizing the need for manual data entry and custom parsing logic. What should you do?

   • ACreate a Cloud Run service that uses the Gemini Pro model to read documents to identify and return the required values in JSON format. Create another Cloud Run service to load the JSON files into a Big Query table. Run the Cloud Run services sequentially.
   • BUse the Cloud Vision API’s OCR feature on a Compute Engine VM to extract all raw text from each document. Write a Python script with regular expressions to search for keywords and parse the extracted text to find the required field values.
   • CSchedule a notebook in an Agent Platform Workbench instance to periodically train a custom named entity recognition (NER) model on a labeled dataset of invoice documents. Deploy the model to an Agent Platform endpoint, and post-process the model output to write results to a BigQuery table.
   • DConfigure a processor using Document AI’s pre-trained invoice processor, and send each document to its endpoint. Extract the structured entities from the processor’s JSON output.

   Show answer

   Answer: D

   Next question

6. Question 336 · 364 in bank

   You developed a demand forecasting model using BigQuery ML. New sales data is ingested into a BigQuery table daily. You want to implement an automated retraining strategy where the model is evaluated against the previous week’s data every Sunday, and the model is retrained if the mean absolute error (MAE) exceeds 500 units. You want to use native BigQuery capabilities to automate this workflow and minimize external dependencies. What should you do?

   • ACreate an Agent Platform Pipelines workflow containing a BigQueryJobOp component that runs the ML.EVALUATE statement. Add a conditional step to check the MAE and conditionally execute a training step. Schedule the pipeline using Agent Platform Pipelines.
   • BWrite a CREATE OR REPLACE MODEL statement with the OPTIONS clause set to warm_start = TRUE. Schedule the query to run every Sunday using a BigQuery scheduled query.
   • CCreate a Cloud Run function triggered by a Cloud Scheduler job. Configure the function to use the BigQuery client library to run the evaluation query. If the MAE threshold is reached, issue a second API call to retrain the model.
   • DUse BigQuery procedural language to write a query that runs the ML.EVALUATE statement, stores the MAE in a variable, and uses an IF statement to conditionally execute the CREATE OR REPLACE MODEL statement. Schedule the script using a BigQuery scheduled query.

   Show answer

   Answer: D

   Next question

7. Question 337 · 364 in bank

   You work in the credit card fraud detection department of a bank that processes millions of transactions per day. You need to create and serve a real-time fraud detection model. To address the challenge of feature engineering at this scale with limited raw data, you plan to create time-related aggregate features, such as the average transaction amount within the last month or the amount of time since the last transaction. How should you process the data for model training and serving?

   • AIngest the transaction data into BigQuery. Use the TRANSFORM clause in BigQuery ML to create new features by applying SQL functions to the raw transaction data.
   • BCreate streaming Dataflow pipelines that read transaction data from Pub/Sub. Use a windowing function to aggregate transaction data over time. Store the engineered features in Agent Platform Feature Store.
   • CIngest the transaction data into BigQuery. Use a TRANSFORM_ONLY model in BigQuery ML to create a reusable feature engineering pipeline. Use this pipeline to preprocess the data.
   • DOrchestrate a workflow in Agent Platform Pipelines that exports raw transaction data from BigQuery to Cloud Storage. Perform feature engineering using a Python script in a custom component in the pipeline.

   Show answer

   Answer: B

   Next question

8. Question 338 · 364 in bank

   You are part of a team of ML engineers building a new churn prediction model using PyTorch. Your team’s raw data is stored in BigQuery. For the initial prototyping phase, your team needs a collaborative, cloud-based environment that meets the following criteria: • It must provide access to GPU accelerators. • It must allow seamless authentication to BigQuery without manual management of service account keys. • It must be pre-configured with common ML frameworks. You need to choose the most secure solution with the least setup overhead for your team to begin model development and experimentation. What should you do?

   • AConfigure a JupyterHub server on a cluster of Compute Engine instances. Install the necessary GPU drivers and other required libraries on each instance. Instruct each team member to create and download a service account key for BigQuery access.
   • BPackage the PyTorch model code in a Docker container, and upload it to Artifact Registry. Use Agent Platform Serverless Training to run a custom training job that runs on a GPU-enabled machine type. Instruct the team to analyze the training logs to iterate on the model.
   • CInstruct team members to use the free version of Agent Platform Colab Enterprise. Use the Agent Platform SDK for Python to load the data into a shared Agent Platform Colab Enterprise notebook. Have each member perform analysis in the shared notebook.
   • DCreate Agent Platform Workbench instances for your team members. Select instance types with attached GPUs and the preinstalled PyTorch kernel. Use the BigQuery client library to access the data, and use the instances’ managed identity for authentication.

   Show answer

   Answer: D

   Next question

9. Question 339 · 364 in bank

   You are responsible for managing the recommendation engine on an ecommerce website. The current model is deployed on a public Agent Platform endpoint and is actively serving production traffic. You have developed an updated version of the model and registered the model in Model Registry. You need to perform an A/B test to compare the new model’s performance to the current model. You plan to implement a traffic split, routing 10% of all incoming production requests to the new model and sending the remaining 90% to the current model. You want this transition to be transparent, and require no modifications to the client application that calls the endpoint. What should you do?

   • ADeploy the new model to a serverless Cloud Run service. Implement logic in the client application that routes 10% of user requests to the Cloud Run service and the rest to the original endpoint.
   • BDeploy the new model to the Agent Platform endpoint where the current model is deployed. Configure the endpoint’s traffic split to allocate 10% of user requests to the new model.
   • CCreate two separate Agent Platform Pipelines, one for each model version. Use Agent Platform Experiments to run both pipelines simultaneously. Configure the experiment to send 10% of user requests to the new model’s pipeline and the rest to the production model’s pipeline.
   • DDeploy the new model to a new Agent Platform endpoint. Configure Cloud Load Balancing to send 10% of user requests to the new endpoint and the rest to the original endpoint.

   Show answer

   Answer: B

   Next question

10. Question 340 · 364 in bank

    You are developing an AI text generator that will be able to dynamically adapt its generated responses to mirror the writing style of the user and mimic famous authors if their style is detected. You have a large dataset of various authors’ works, and you plan to host the model on a custom VM. You want to use the most effective model. What should you do?

    • AFine-tune the latest Llama model from Model Garden on Agent Platform Serverless Training.
    • BUse the latest Gemini Flash foundational model.
    • CDeploy the latest Llama model from Model Garden, and use prompt engineering.
    • DFine-tune a BERT-based model from Hugging Face.

    Show answer

    Answer: A

    Next question

Keyboard:J Kmove between questionsA–Dselect an answerRshow or hide the answer← →change page

PMLE practice, question 331. 364 questions available.