Databricks-ML-Pro ML Ops Practice Test 5 — 15 Questions

Question 1

Which TWO statements regarding the use of Unity Catalog in Databricks for MLOps are correct?

Accepted Answer

It enables centralized lineage tracking between data tables and MLflow models.. Unity Catalog acts as the centralized governance layer for all data and AI assets. It provides fine-grained access control and lineage tracking, allowing organizations to maintain visibility over which data is used by which models. By centralizing these assets, teams can collaborate safely while ensuring that compliance and auditing requirements are satisfied across the entire organization, regardless of the individual workspace where the work is performed.

Answer

Unity Catalog enforces access control at the table level only, not on MLflow models.

Answer

Unity Catalog is only compatible with Python and does not support SQL-based model management.

Answer

Users must manually export model artifacts to an external store to share them via Unity Catalog.

Question 2

You are designing a model retraining strategy. What is the most reliable way to trigger a retraining job based on model performance degradation?

Accepted Answer

Use Databricks Workflows to trigger retraining based on drift detection alerts.. Linking monitoring metrics to workflow orchestration is the standard MLOps pattern for automated retraining. By configuring a monitor that triggers a Databricks Job when performance drifts below a defined metric (e.g., accuracy), you create a closed-loop system. This ensures that models are updated only when necessary, minimizing cost while maintaining high model performance and reducing manual intervention in the model lifecycle.

Answer

Schedule the retraining job to run every Monday regardless of current model performance.

Answer

Send an email alert to the lead data scientist to manually kick off the job.

Answer

Write a cron job in the notebook that calculates performance every minute.

Question 3

Refer to the exhibit. The deployment pipeline is failing to load the model artifact in the target production environment. What is the most likely cause?

Accepted Answer

The path '/dbfs/...' is not a reliable way to reference artifacts across different clusters.. The exhibit shows a hard-coded path beginning with '/dbfs/'. In Databricks, accessing DBFS via local file system paths can be inconsistent across different clusters or environments. The correct approach is to use the MLflow URI (e.g., 'runs:/...') to load models, as it abstracts the underlying storage location and ensures the artifact is resolved correctly regardless of the execution environment or storage configuration.

Answer

The model version is not set to 'PRODUCTION' in the metadata.

Answer

The conda_env file is missing from the artifact storage.

Answer

The run_id 'd7a8e9f0' has been archived and is no longer accessible.

Question 4

Which component of Databricks is specifically designed to prevent training-serving skew by ensuring that feature engineering code is consistent during both model training and real-time inference?

Accepted Answer

Databricks Feature Store. The Databricks Feature Store acts as a centralized repository for features. By using the same feature definitions and transformation logic for both training and serving, it eliminates the discrepancy known as training-serving skew. This ensures that the features fed to the model at inference time are calculated exactly as they were during training, maintaining consistent performance and model reliability.

Answer

MLflow Tracking Server

Answer

Unity Catalog

Answer

Databricks SQL

Question 5

You are implementing a CI/CD pipeline for a model. You want to automate unit testing of the model's inference performance. Which approach is best suited for Databricks?

Accepted Answer

Trigger a notebook to run inference on a validation set and assert metrics.. Using a dedicated job workflow that triggers a notebook to run inference on a hold-out test dataset is the standard Databricks approach. By comparing metrics against defined thresholds before deployment, you ensure quality. This automated gate prevents poor-performing models from reaching production, which is a critical step in a mature MLOps lifecycle to maintain system stability and model accuracy.

Answer

Manually inspect performance metrics in MLflow UI before every deployment.

Answer

Deploy the model to staging and wait for user feedback.

Answer

Rely on the Model Registry to automatically validate model performance.

Question 6

Refer to the exhibit. If the 'train_model' task fails, what happens to the 'evaluate_model' task in this Databricks Job?

Accepted Answer

It is skipped because the dependency is not met.. In Databricks Jobs, a task that depends on another will only execute if the parent task finishes successfully. If 'train_model' fails, the dependency condition for 'evaluate_model' is not met, causing the job to stop or skip the downstream task. This behavior is intentional to prevent evaluating a non-existent or corrupted model, ensuring pipeline integrity and avoiding wasted computational resources on dependent tasks.

Answer

It executes regardless of the failure.

Answer

It attempts to run using the last successful model artifact.

Answer

It retries the 'train_model' task automatically until success.

Question 7

Your organization wants to monitor production models for drift. Which Databricks service should be used to detect changes in the input data distribution compared to the training data?

Accepted Answer

Databricks Lakehouse Monitoring. Databricks Lakehouse Monitoring is the native solution for tracking data quality and drift. By analyzing incoming data against a baseline established during training, it identifies statistical deviations. This is critical for MLOps because detecting drift early allows data scientists to trigger retraining or investigate data pipeline issues, ensuring model performance remains consistent over time despite changing real-world data patterns.

Answer

MLflow Model Registry

Answer

Delta Live Tables (DLT)

Answer

Databricks SQL Alerting

Question 8

Which action should be taken to ensure that sensitive PII data is not leaked when logging models using MLflow?

Accepted Answer

Ensure data used for model artifacts is scrubbed of PII.. Sanitizing data before logging or excluding sensitive columns from the training set is a security best practice. MLflow logs metadata and often sample data. If PII is included, it is persisted in the artifact store. By ensuring only non-sensitive features are logged, you comply with data privacy regulations and prevent unauthorized access to sensitive information within your model artifacts.

Answer

Encrypt the entire MLflow experiment folder using workspace-level encryption.

Answer

Set the MLflow model registry to 'Private' mode.

Answer

Delete the MLflow run immediately after deployment.

Question 9

In the context of Databricks MLOps, what is the primary purpose of a 'Staging' environment in the Model Registry?

Accepted Answer

To allow for final validation and testing before production deployment.. The Staging environment serves as a pre-production gate where models undergo final validation, such as integration testing and UAT (User Acceptance Testing), before being promoted to Production. This ensures that only models that meet performance, safety, and operational standards are exposed to live production traffic, reducing the risk of catastrophic failures in the real-world deployment environment.

Answer

To act as a long-term backup for model version history.

Answer

To automatically convert models into a different serving format.

Answer

To increase the model's training accuracy.

Question 10

When deploying a model to a production environment, why is it recommended to use a 'Model Signature'?

Accepted Answer

To define the input and output schema for automatic validation.. A model signature defines the schema of the inputs and outputs. By enforcing this, the serving endpoint can automatically validate incoming requests and reject those that do not match the expected format. This prevents runtime errors and unexpected behavior, serving as a critical safety feature in automated production pipelines where data quality might fluctuate, ensuring the model always receives the correct data structure.

Answer

To increase the inference speed of the model.

Answer

To automatically retrain the model if the input data changes.

Answer

To encrypt the model artifacts in the registry.

Question 11

Which TWO of the following are primary benefits of using the MLflow Model Registry in Databricks?

Accepted Answer

Lifecycle state tracking (e.g., Staging, Production).. The Model Registry centralizes model management, providing a single source of truth for model lifecycle stages and version history. This is vital for MLOps, as it ensures that teams can track which models are in production, who approved them, and how they perform over time, enabling consistent deployment workflows and improved auditability for regulatory compliance and internal governance.

Answer

Automated training of models.

Answer

Real-time monitoring of inference latency.

Answer

Automatic data cleaning for training sets.

Question 12

Which of the following describes the 'Gold' layer in the Medallion Architecture, and why is it important for machine learning?

Accepted Answer

It contains aggregated, business-level data ready for model training.. The Gold layer contains refined, business-level data that is ready for consumption. In MLOps, it is the standard source for training data. By using curated, high-quality data from the Gold layer, data scientists avoid the noise and inconsistencies of raw data, leading to more robust models and faster iteration times, as they spend less time on manual data cleaning and validation.

Answer

It contains raw data ingested directly from external sources.

Answer

It is used to store model artifacts and logs.

Answer

It is where data scientists perform exploratory data analysis.

Question 13

Refer to the exhibit. A data scientist attempted to register a new model version but received the error shown in the exhibit. Which step should be taken to resolve this issue?

Accepted Answer

Ask an administrator to grant 'CAN_MANAGE' or 'CAN_EDIT' permissions on the model in the MLflow UI.. The error indicates a lack of necessary permissions to perform operations on the MLflow Model Registry in the Databricks workspace. Access control in Databricks is granular; even if a user can write code, they need specific registry-level permissions to promote or register models. This is a common MLOps governance issue where security policies must be correctly balanced with developer autonomy.

Answer

Upgrade the user's Databricks entitlement to 'Workspace Admin'.

Answer

Update the MLflow experiment tracking URI to point to an external database.

Answer

Re-run the training notebook using the cluster owner's credentials.

Question 14

Which feature in Databricks allows you to automatically track training code, parameters, metrics, and models during the development phase?

Accepted Answer

MLflow Tracking. MLflow Tracking is the core component of Databricks for experiment management. It provides a structured way to log and visualize the results of machine learning runs. Understanding this component is fundamental to MLOps because it allows teams to compare models, track performance over time, and ensure that every experiment is reproducible, which is essential for professional model development and auditing.

Answer

Databricks Delta Lake

Answer

Databricks Feature Store

Answer

Databricks Workflows

Question 15

Your organization requires that all models deployed to production must be signed by a security officer. How can you enforce this requirement within the Databricks MLflow Model Registry?

Accepted Answer

Restrict 'CAN_MANAGE' permissions on the model, and use an automated CI/CD pipeline for promotion.. Enforcing approval workflows is crucial for compliance and risk management in MLOps. By using MLflow's permissions and stages, you can restrict who can promote models to 'Production'. Combining this with programmatic checks or external CI/CD gates ensures that a model cannot be deployed without the necessary authorization, preventing unauthorized or unvetted code from reaching production inference services.

Answer

Delete all models in the Production stage and only re-upload the signed models.

Answer

Grant 'CAN_MANAGE' permissions on the registry to everyone to ensure transparency.

Answer

Set the model version description to 'Signed by Security' after manual inspection.