Databricks-ML-Pro Model Development Practice Test 6 — 15 Questions

Question 1

A machine learning engineer wants to ensure that model training artifacts are persistent and accessible even if the ephemeral compute cluster is terminated. What is the standard practice in Databricks for achieving this?

Choose an answer to begin — your selection is scored in the full session.

Accepted Answer

Log the model using MLflow, which stores the artifact in the configured backend.. In Databricks, local cluster storage is ephemeral and is deleted when the cluster terminates. To ensure model artifacts persist, developers must log them to MLflow, which automatically handles the backend storage in DBFS or Unity Catalog-managed storage. This practice decouples the model lifecycle from the compute lifecycle, ensuring that models remain available for deployment or further evaluation after the compute resources are released.

Answer

Save the model to the local file system using the /dbfs/ path prefix.

Answer

Export the model artifact to a temporary workspace file and download it locally.

Answer

Configure the cluster to use an external persistent volume for local storage.

Question 2

When developing a model, which THREE actions should a data scientist perform to ensure the model is ready for production deployment via Model Serving?

Accepted Answer

Log the model with a defined input/output signature.. Production-ready models require strict standards: a defined schema (signature) for input validation, proper metadata logging for reproducibility, and registration in the Model Registry. These steps ensure the model is governable, traceable, and secure. Skipping any of these components makes the model difficult to debug, deploy reliably, or monitor for performance drift in production environments.

Answer

Hardcode the data path to an external S3 bucket within the model object.

Answer

Include the entire training dataset in the model's metadata.

Question 3

Refer to the exhibit. A developer wants to ensure the Random Forest model can be used for automated inference at scale. Based on the provided code, what is missing to enable the model to support the 'predict' method within the Databricks Model Serving environment?

Accepted Answer

The model.fit(X_train, y_train) method must be called before log_model.. The provided code logs an unfitted model instance. MLflow cannot serialize a model's 'predict' capabilities until the model has actually been trained (fitted) on a dataset. Attempting to deploy or run an unfitted model results in an error, as there is no learned logic to execute. The model must be fitted with data prior to being passed to log_model.

Answer

The model must be registered to the model registry before logging.

Answer

The model must be wrapped in an mlflow.pyfunc.PythonModel class.

Answer

The code must explicitly define the input signature in log_model.

Question 4

When using the Databricks Model Registry, what does a 'Model Version' represent in the context of the lifecycle?

Accepted Answer

A unique, immutable snapshot of a specific model artifact and its metadata.. A Model Version is a point-in-time snapshot of a model, including the code, model weights, dependencies, and environment configuration. By tracking these versions, data scientists can compare performance across different iterations, rollback to previous states if a production deployment fails, and maintain an audit trail of how a model evolved over time. This versioning is foundational for reproducible machine learning.

Answer

An automated report generated after every training run.

Answer

A live pointer that always points to the most recent training run.

Answer

A shared directory containing the raw training data and configuration files.

Question 5

Which method is the most appropriate for logging custom pre-processing logic alongside a model so that it is automatically applied during inference in Databricks?

Accepted Answer

Use the mlflow.pyfunc flavor to wrap the model and transformation logic.. The mlflow.pyfunc flavor is designed specifically for this purpose. It allows developers to define a custom class that inherits from PythonModel, where both the pre-processing logic and the model prediction logic reside in the same artifact. This ensures that the exact same transformation steps applied during training are applied during inference, preventing training-serving skew, which is a common source of production model degradation.

Answer

Include pre-processing in the notebook and rely on the serving user to call it.

Answer

Store the transformation logic in a separate repository and import it at inference time.

Answer

Write the pre-processing code as a SQL view in the Databricks SQL Warehouse.

Question 6

When using MLflow to manage the machine learning lifecycle, what is the primary purpose of the 'conda.yaml' or 'requirements.txt' file automatically generated during log_model?

Accepted Answer

To ensure that the inference environment has the necessary dependencies installed.. These files define the environment specification required to recreate the model's runtime environment. When a model is moved to a production serving endpoint or a different cluster, Databricks uses these specifications to install the correct library versions. This ensures that the model executes in an environment identical to the one it was trained in, preventing silent failures caused by library version mismatches.

Answer

To store the model's hyperparameter search space configurations.

Answer

To act as a security manifest that validates the digital signature of the model.

Answer

To limit the maximum number of concurrent requests the model can handle.

Question 7

A data scientist is iterating on a model and notices that their training runs are becoming disorganized. What is the standard Databricks mechanism for tracking different 'attempts' at model improvement within a single project?

Accepted Answer

Grouping runs within an experiment and using MLflow tags.. MLflow tracking allows for multiple 'runs' under a single experiment. Each run captures parameters, metrics, and artifacts. By systematically naming or tagging these runs, the data scientist can easily compare performance metrics (like RMSE or F1-score) in the MLflow UI. This structured approach is fundamental to scientific experimentation and prevents the loss of valuable insights when iterating on model architectures or hyperparameter sets.

Answer

Using a new Git branch for every single hyperparameter change.

Answer

Writing the results to a shared CSV file in the root directory.

Answer

Storing all run parameters in the notebook's global variable state.

Question 8

When working in Databricks, where should a data scientist primarily look to monitor the resource utilization and execution logs of an active model training job?

Accepted Answer

The Databricks Jobs UI.. The 'Jobs' UI and the Spark UI are the primary interfaces for monitoring training jobs. While MLflow provides tracking for metrics and parameters, the Jobs UI provides the necessary system-level visibility into cluster performance, execution status, and logs. This is critical for troubleshooting memory errors, diagnosing bottlenecks, and managing costs during resource-intensive training sessions.

Answer

The MLflow Model Registry.

Answer

The Databricks Filesystem (DBFS) root directory.

Answer

The Workspace 'Notebook' sidebar.

Question 9

Refer to the exhibit. A user wants to retrieve the 'accuracy' metric from this run programmatically. Which code snippet correctly accesses this value?

Accepted Answer

client = mlflow.tracking.MlflowClient(); client.get_run('550e8400-e29b-41d4-a716-446655440000').data.metrics['accuracy']. The MLflow tracking client provides the `get_run` method, which returns an object containing all run metadata, including metrics. Accessing the dictionary via `data.metrics` is the standard way to retrieve tracked values. This is critical for automated model comparison scripts, where developers must programmatically evaluate multiple runs to select the best-performing iteration for registration in the Model Registry.

Answer

mlflow.get_metric('accuracy', run_id='550e8400-e29b-41d4-a716-446655440000')

Answer

mlflow.search_runs(run_ids=['550e8400-e29b-41d4-a716-446655440000'])['accuracy']

Answer

mlflow.load_metric('accuracy', '550e8400-e29b-41d4-a716-446655440000')

Question 10

When utilizing the Databricks Feature Store for model development, why should a developer define a primary key in the Feature Table?

Accepted Answer

To enable the Feature Store to perform automated lookups for model inference.. Defining a primary key is essential for enabling point-in-time joins during model inference. It allows the Feature Store to perform lookups efficiently and ensures that data consistency is maintained between training and serving. By using primary keys, the system can automatically handle complex join operations, reducing the risk of data leakage and simplifying the feature engineering pipeline for production deployment.

Answer

To increase the write throughput of the Delta table underlying the feature store.

Answer

To bypass the need for performing data validation checks on features.

Answer

To automatically convert all features into a sparse matrix format.

Question 11

A team is developing a model to forecast demand. They need to ensure that their feature engineering code is reusable for both training and real-time inference. Which architectural pattern should they adopt?

Accepted Answer

Create a Feature Store table and register feature engineering as a Feature Spec.. Defining features within the Databricks Feature Store allows the same code to be used for batch training and real-time lookup. This eliminates the risk of training-serving skew, a major cause of production model degradation. By centralizing feature engineering, teams ensure that the transformations applied during training are identical to those applied during inference, maintaining consistency across the entire development and deployment pipeline.

Answer

Store preprocessing logic in individual notebook files and import them dynamically.

Answer

Perform all feature engineering inside the model's 'predict' method.

Answer

Write all features to a CSV file and load them during inference.

Question 12

When hyperparameter tuning using `mlflow.spark.autolog()` or `hyperopt`, what is the primary advantage of logging the parameters to the MLflow tracking server?

Accepted Answer

It allows developers to visualize and compare results across hundreds of tuning iterations.. Logging parameters allows for comprehensive lineage and comparison of model runs. It enables the 'experimentation' phase to be evidence-based, where developers can correlate parameter configurations with model performance metrics. This is crucial for reproducibility and for identifying the optimal configuration that maximizes model performance, as it creates a permanent, queryable history of the entire tuning process.

Answer

It automatically triggers a new training run when the parameters are updated.

Answer

It compresses the model weights to reduce the storage footprint on DBFS.

Answer

It prevents overfitting by forcing the model to use regularization parameters.

Question 13

A data scientist is training a model on a large Delta table. They want to ensure that the training data remains consistent even if the underlying table is updated during the training process. What is the most robust way to achieve this?

Accepted Answer

Use Delta Lake Time Travel to query the table at a specific version.. Delta Lake's time travel feature allows users to query a specific version or timestamp of a table. By using `VERSION AS OF` or `TIMESTAMP AS OF` syntax, the training job can anchor itself to an immutable snapshot of the data. This is critical for model reproducibility, as it ensures that the training set does not change mid-experiment, preventing non-deterministic results and enabling accurate auditing.

Answer

Create a temporary local copy of the table in the driver's memory.

Answer

Lock the table using a SQL 'LOCK TABLE' statement during the job.

Answer

Manually filter data by a 'processed_timestamp' column in every query.

Question 14

Refer to the exhibit. What is the purpose of the 'signature' section in this model configuration?

Accepted Answer

To ensure that inference requests match the expected data types and structure.. The signature defines the schema of the model's inputs and outputs. This metadata is used by MLflow and model serving infrastructure to validate inference requests before they reach the model. By enforcing types and expected fields, the system prevents runtime errors due to malformed input, which is critical for maintaining robust production inference pipelines in distributed environments.

Answer

To encrypt the model artifacts before they are saved to the registry.

Answer

To specify the optimization level for the model training process.

Answer

To define the hyperparameters used for the model's final evaluation.

Question 15

Which THREE features are provided by the Databricks Model Registry for model lifecycle management?

Accepted Answer

Transitioning models between stages (e.g., Staging to Production).. The Model Registry provides the governance layer for machine learning, allowing teams to track stages, version history, and approval workflows. These features are necessary to transition from an experimental notebook-based workflow to a production environment where changes to models are managed, tested, and audited, ensuring that only validated artifacts are promoted to critical business-facing applications.

Answer

Automatic retraining of models on a fixed daily schedule.

Answer

Automated deployment of models to edge devices like mobile phones.