Databricks-ML-Pro ML Ops Practice Test 7 — 15 Questions

Question 1

Your team is migrating models to Databricks Model Registry. You need to ensure that production models are only deployed after passing specific validation tests. Which feature best facilitates this automated governance?

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

Accepted Answer

Configuring Model Registry webhooks for status transitions.. Model Registry webhooks are the critical component for automating governance. By triggering external CI/CD pipelines (like Jenkins or GitHub Actions) upon model version transitions (e.g., 'Pending' to 'Production'), organizations enforce quality gates. This ensures that only models validated by automated unit tests or performance benchmarks are promoted, reducing manual intervention and minimizing the risk of deploying underperforming models into production environments.

Answer

Using MLflow Experiments to log model artifacts.

Answer

Hard-coding model versions in inference notebooks.

Answer

Executing manual SQL queries against the MLflow backend.

Question 2

You are auditing a Databricks environment to ensure compliance. Which TWO actions ensure the highest level of model lineage and reproducibility for models registered in MLflow?

Accepted Answer

Include the Git hash in the log_model metadata.. Linking models to specific Git commits and tracking the environment dependencies via Conda or pip requirements files are essential for reproducibility. These steps ensure that when a model is redeployed, the exact code version and environment can be recreated, fulfilling the core MLOps requirement of auditability and consistent model behavior across different computing environments.

Answer

Use random seeds for all ML libraries.

Answer

Store all training data as CSV files in the workspace.

Answer

Manually copy artifacts to a shared folder.

Question 3

Your organization requires a feature store strategy that supports low-latency point-in-time lookups for online inference. Which implementation approach best addresses this?

Accepted Answer

Utilizing the Databricks Feature Store with online store integration.. Databricks Feature Store enables point-in-time joins to avoid data leakage during training and supports publishing to low-latency online stores (e.g., DynamoDB or Cosmos DB). By using a centralized feature store, teams ensure consistency between training and serving. Point-in-time lookups are crucial to ensure that features used for predictions reflect exactly what was known at that specific moment, preventing bias.

Answer

Using a temporary Spark SQL view for features.

Answer

Caching all feature data in local cluster memory.

Answer

Performing manual joins in inference notebooks.

Question 4

Which component of MLflow tracks the parameters, metrics, and tags associated with a specific training run?

Accepted Answer

MLflow Tracking.. MLflow Tracking is the primary API used for logging and querying machine learning experiments. It allows data scientists to record key information during the training phase, which is vital for experiment reproducibility and comparing performance across different model iterations. Without MLflow Tracking, maintaining visibility into the history of experiments would be manual, fragmented, and prone to losing critical context.

Answer

MLflow Model Registry.

Answer

MLflow Projects.

Answer

MLflow Recipes.

Question 5

Your ML pipeline requires a complex environment with specific C++ dependencies. What is the recommended way to manage this in Databricks?

Accepted Answer

Use Databricks Container Services with a custom image.. Databricks Container Services allows users to define custom Docker images as the execution environment. This is the only way to handle complex system-level dependencies like C++ libraries that cannot be installed via standard pip or conda commands. By using custom containers, teams ensure that the training and inference environments are identical and include all necessary system-level components.

Answer

Install dependencies in a setup script at runtime.

Answer

Include dependencies in the MLflow model artifact.

Answer

Create a shared library JAR file.

Question 6

Refer to the exhibit. Which security configuration is the likely culprit for this job failure?

Accepted Answer

Insufficient Unity Catalog permissions for the job identity.. This error points to a failure in Unity Catalog access control. In Databricks, when a job attempts to read from a table, it requires specific 'SELECT' permissions on the table and 'USE' permissions on the catalog and schema. If the job service principal is not granted these roles, the query will fail regardless of the workspace-level permissions assigned to the user.

Answer

Missing cluster-level library permissions.

Answer

The table is stored in an encrypted S3 bucket.

Answer

The Delta table has reached its version limit.

Question 7

A team wants to track model performance over time to detect drift without writing custom monitoring infrastructure. What is the most efficient Databricks tool for this?

Accepted Answer

Databricks Model Monitoring (DMM).. Databricks Model Monitoring (DMM) provides built-in capabilities to track model performance, data drift, and input quality. By automating these tasks, teams avoid the technical debt of building and maintaining custom monitoring solutions. DMM integrates seamlessly with the Model Registry, allowing for automated alerts and reports that keep stakeholders informed about model health in production environments.

Answer

MLflow Tracking APIs.

Answer

Delta Lake Change Data Feed.

Answer

Databricks SQL Dashboards.

Question 8

When deploying a model to a production endpoint, your team requires that the model is served from a specific, immutable version. How do you ensure this in MLflow?

Accepted Answer

Specify the exact model version in the serving configuration.. Linking the deployment to a specific Model Registry version alias or version number is the only way to ensure immutability. When you serve a model, you should point to a specific version (e.g., version 5) rather than a dynamic alias like 'Production' if you need to guarantee that no subsequent transitions will change the behavior of the endpoint, thereby ensuring production stability.

Answer

Always point the endpoint to the 'Production' tag.

Answer

Use the latest model artifact from the experiment folder.

Answer

Rebuild the model container on every request.

Question 9

Which Databricks feature is specifically designed to prevent data leakage during model training by ensuring feature values are fetched as they existed at a specific point in time?

Accepted Answer

Databricks Feature Store.. Point-in-time joins are the core mechanism of the Databricks Feature Store. When training on historical data, it is critical to use features that were available at that time, not their current values. This prevents 'look-ahead bias,' where future information inadvertently leaks into the training set, causing the model to perform artificially well during training but failing in production.

Answer

Delta Lake Time Travel.

Answer

MLflow Model Registry.

Answer

Auto Loader.

Question 10

In an automated MLOps workflow, what is the best practice for handling model training failures?

Accepted Answer

Logging the error and triggering an automated notification.. Failing fast and sending alerts is the standard practice. Automated pipelines should include error handling that logs the failure cause to MLflow and triggers a notification via email or Slack. This ensures that data scientists can address the issue immediately without waiting for a manual audit of the pipeline, maintaining the overall health and reliability of the automated machine learning lifecycle.

Answer

Retrying the training job indefinitely.

Answer

Ignoring the error and deploying the previous model.

Answer

Deleting the failed experiment run.

Question 11

Your team uses Databricks Model Serving to host a production model. You need to ensure zero-downtime updates while maintaining the ability to revert to the previous version instantly if performance degrades. Which deployment strategy should you implement?

Accepted Answer

Use a Blue-Green deployment strategy with traffic splitting.. Blue-green deployment allows you to maintain a production endpoint while simultaneously deploying a new version. By using weighted traffic splitting, you can shift traffic gradually and monitor performance metrics. If the new version exhibits anomalies, you can instantly revert traffic to the stable version, ensuring high availability and minimizing the impact of potential production failures, which is critical for robust MLOps workflows.

Answer

Directly update the existing model serving endpoint to the new version.

Answer

Delete the current serving endpoint and create a new one with the updated model.

Answer

Increase the instance count of the existing endpoint before updating.

Question 12

Which TWO actions are required to properly implement MLflow Model Registry stages and governance for a machine learning project?

Accepted Answer

Transition model versions between stages like 'Staging' and 'Production'.. Implementing Model Registry stages ensures that only validated models move from development to production. By controlling access permissions, organizations enforce a separation of duties, ensuring that data scientists can register models, while only authorized engineers can promote them. This gated workflow is foundational for regulatory compliance and auditability in enterprise MLOps, preventing unauthorized model deployments into downstream production environments.

Answer

Assign the 'Can Manage' permission on the Model Registry to all users.

Answer

Keep all model versions in the 'None' stage for maximum flexibility.

Answer

Hard-code the model version string into the production application code.

Question 13

Refer to the exhibit. Your automated CI/CD pipeline triggered a model deployment to production, but the job failed with the error shown. What is the most likely cause?

Accepted Answer

The deployment job is referencing a model version that hasn't been created yet.. The error indicates that the deployment pipeline is attempting to access a model version that does not exist in the Model Registry. This usually happens when the pipeline triggers before the registration process completes or when a race condition occurs between the training job and the deployment job. Validating the existence of the model version before initiating deployment is essential for pipeline reliability.

Answer

The model was registered, but the cluster lacks permissions to read it.

Answer

The model serving endpoint is already occupied by another model version.

Answer

The training cluster ran out of memory during the model artifact upload.

Question 14

When implementing a Feature Store in Databricks, what is the primary benefit of using a Feature Table compared to a standard Delta table for feature engineering?

Accepted Answer

They enable automated point-in-time joins for training data generation.. Feature Tables provide metadata tracking and lineage, linking features directly to the models that consume them. This metadata enables automated point-in-time joins, preventing data leakage during training by ensuring features are sampled as they existed at a specific timestamp. This level of automation is not available with standard Delta tables, making Feature Tables essential for building reproducible, production-ready machine learning pipelines.

Answer

Feature Tables automatically convert all data into Parquet format.

Answer

They provide faster query performance than standard Delta tables.

Answer

They allow users to write SQL queries directly to the feature data.

Question 15

Your team is experiencing 'training-serving skew' where model performance in production is significantly lower than during training. Which approach should you prioritize to mitigate this issue?

Accepted Answer

Use the Databricks Feature Store to unify feature engineering pipelines.. Training-serving skew often occurs because the data transformation logic used during model training differs from the logic applied during real-time inference. By utilizing the Databricks Feature Store, you can centralize your feature engineering code. This ensures that the exact same transformation code and look-up logic are utilized for both offline training and online inference, effectively eliminating the discrepancy between the two environments.

Answer

Increase the amount of historical training data used in the model.

Answer

Switch the model architecture to a deeper neural network.

Answer

Retrain the model every hour to keep it fresh with recent data.