Databricks-ML-Pro Practice Questions 2 — 50 Questions

Question 1

You are designing a CI/CD pipeline for a machine learning model on Databricks. The pipeline must automatically retrain the model when new data arrives, validate it, and deploy it to a serving endpoint if it passes. Which two components are essential to achieve this? (Choose two.)

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

Accepted Answer

A Databricks Model Serving endpoint that can be updated with the new model version via the MLflow API.. A Databricks Job provides the orchestration to run training and validation when new data arrives, and a Model Serving endpoint is required to deploy the validated model for real-time predictions. Together, they form the backbone of an automated retraining and deployment pipeline. MLflow Tracking, feature stores, and Repos are valuable but not strictly essential for the pipeline to function.

Answer

MLflow Tracking to log parameters, metrics, and artifacts for each run.

Answer

A Databricks Repos integration to version control notebooks and manage code changes.

Answer

A feature store that automatically ingests new data and updates features in real time.

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

A data scientist is using MLflow to track experiments on Databricks. They notice that some runs are missing the model artifact even though they called mlflow.sklearn.log_model(). What is the most likely cause?

Accepted Answer

The model artifact was logged outside of an active MLflow run.. MLflow only logs artifacts when there is an active run. Calling mlflow.sklearn.log_model() outside a run context results in the artifact not being associated with any run, so it appears missing. Other options like size limits or overwrites are less likely. Ensuring the logging call is inside a with mlflow.start_run() block or after mlflow.start_run() will correctly attach the model artifact to the run.

Answer

The model artifact was overwritten by a subsequent run with the same name.

Answer

The model artifact was too large and exceeded the maximum artifact size.

Answer

The model artifact was logged to a different experiment than the one being viewed.

Question 4

You are retiring a real-time model endpoint on Databricks Model Serving. The endpoint has been serving production traffic for six months and you want to archive its request logs for compliance before deleting the endpoint. Which action should you take first?

Accepted Answer

Query the existing inference table that was configured for the endpoint and export its contents before deleting the endpoint.. Model Serving writes inference payloads to a Delta inference table when the feature is enabled, and that table survives endpoint deletion because it is a separate Unity Catalog object. Exporting the table before removing the endpoint preserves the historical request and response records required for compliance, whereas configuration snapshots or model artifacts do not contain traffic data.

Answer

Download the endpoint's model artifact from the MLflow Model Registry and store it as the audit record.

Answer

Use the Databricks REST API to list the endpoint's configuration and save the JSON response as the archive.

Answer

Enable inference table logging on the endpoint and wait for the next inference requests to be persisted.

Question 5

Refer to the exhibit. An engineer is configuring a serving endpoint. Based on the configuration provided, what is the impact of the 'auto_scale' flag?

Accepted Answer

It allows the serving endpoint to adjust worker count based on request load.. The 'auto_scale' flag allows the Databricks Model Serving endpoint to dynamically adjust the number of active workers based on real-time traffic load. This ensures that the system maintains performance during peak request periods while minimizing costs during low-traffic periods. This balance of performance and efficiency is a critical aspect of managing production ML infrastructure in an enterprise Databricks environment.

Answer

It scales the number of model versions deployed in the registry.

Answer

It automatically re-trains the model when request volume drops.

Answer

It forces the endpoint to only use 10 workers permanently.

Question 6

Which of the following is an advantage of using Databricks AutoML compared to building a custom Scikit-Learn training loop?

Accepted Answer

It generates the training notebook for the trial, allowing for further refinement and transparency.. Databricks AutoML significantly accelerates the development lifecycle by automating tedious tasks like data cleaning, feature engineering, and model selection. It provides a baseline of high-performance models while simultaneously producing the source code for the best-performing model. This allows teams to iterate rapidly, then customize the generated code, combining the speed of automation with the flexibility of manual tuning for complex production requirements.

Answer

AutoML models are guaranteed to have higher accuracy than custom models.

Answer

It prevents the need for any feature engineering by automatically creating all required variables.

Answer

It eliminates the need for monitoring the model after deployment.

Question 7

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.

Question 8

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 9

An ML engineer is deploying a model to Databricks Model Serving and wants to implement A/B testing between two model versions. The engineer needs to route a percentage of traffic to each version and collect performance metrics. Which feature of Databricks Model Serving should the engineer use?

Accepted Answer

Create a single endpoint with multiple model versions and configure traffic splitting between them.. To perform A/B testing between model versions, the engineer should create a single Model Serving endpoint that serves multiple model versions with traffic splitting. This allows a specified percentage of requests to be routed to each version, and Databricks provides metrics for each version, facilitating performance comparison. This native feature simplifies A/B testing.

Answer

Deploy two separate endpoints and use an external load balancer to distribute traffic between them.

Answer

Enable the 'Canary' deployment option in the endpoint configuration to automatically split traffic.

Answer

Use MLflow's model registry to assign a stage to each model version and route traffic based on the stage.

Question 10

Refer to the exhibit. What happens to these logged metrics in MLflow when the training run completes?

Accepted Answer

MLflow keeps all values, enabling the visualization of metrics over time.. MLflow is designed to track metrics over time, which is crucial for monitoring model convergence during training. By logging the same metric key multiple times, MLflow stores each value as a point in a time series. This allows developers to visualize the training progress and identify when the model stops improving, which is critical for implementing early stopping and optimizing training efficiency in deep learning.

Answer

MLflow overwrites the previous value with the latest value.

Answer

The run will throw an error because the metric key is not unique.

Answer

MLflow averages the values and stores the mean.

Question 11

A machine learning engineer is using MLflow Tracking to log metrics and artifacts for a deep learning model. They notice that the training run logs a large number of metrics (e.g., loss per batch) and want to reduce the storage footprint and improve query performance. They also need to retain the ability to compare runs and reproduce results. Which of the following actions is most appropriate?

Accepted Answer

Log only the final epoch metrics and use MLflow's step parameter to log intermediate metrics less frequently.. MLflow's log_metric function can accept a step parameter to record metrics over time, but logging every batch creates a large number of entries. Reducing the logging frequency or only logging final epoch metrics significantly decreases storage and improves query speed while still providing enough data for comparison and reproducibility. Other options either lose MLflow's tracking capabilities or introduce unsupported configurations.

Answer

Disable metric logging entirely and rely on TensorBoard for visualization.

Answer

Use a different tracking server with a NoSQL backend to handle the high volume of metrics.

Answer

Store metrics in a separate file and log it as an artifact instead of using mlflow.log_metric.

Question 12

A machine learning engineer is using MLflow to track experiments and has logged a model with a signature. They now want to register this model in the MLflow Model Registry and promote it to Production. Which MLflow API call should they use to add the model to the registry?

Accepted Answer

mlflow.register_model(model_uri, name). The mlflow.register_model function is designed to take an existing model URI and add it as a new version to the Model Registry. It is the standard way to register a model after logging. Other options either create empty models or log new models, which do not fit the scenario of promoting an already logged model.

Answer

mlflow.log_model(model, artifact_path, registered_model_name=name)

Answer

mlflow.pyfunc.save_model(path, loader_module, data_path, registered_model_name=name)

Answer

mlflow.create_registered_model(name)

Question 13

A machine learning team is transitioning from local notebooks to Databricks. They want to ensure their code is modular and reusable. Which THREE practices should they implement?

Accepted Answer

Package common utility code into Python wheels.. Transitioning to Databricks requires moving away from monolithic notebooks toward modular code structures. Using Delta Lake for data consistency, moving logic into Python wheels, and utilizing Repos for version control are the standard industry practices. These steps ensure that code is maintainable, testable, and capable of being integrated into automated CI/CD pipelines, which is the hallmark of mature MLOps practices within a Databricks workspace.

Answer

Keep all training, preprocessing, and evaluation code in a single notebook.

Answer

Hardcode all file paths and credentials in every notebook.

Question 14

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 15

A team is building a feature store in Databricks. They need to ensure that training data and inference data are consistent to avoid training-serving skew. What is the primary benefit of using the Databricks Feature Store in this context?

Accepted Answer

It provides a unified interface for computing features, ensuring consistency between training and inference.. The Databricks Feature Store enables the reuse of feature pipelines, ensuring that the exact same transformation logic is applied during both training and inference. By decoupling feature generation from model training, it creates a 'single source of truth' for features, effectively eliminating training-serving skew and improving the reliability of models in production. This is a vital component for maintaining feature consistency across distributed teams.

Answer

It automatically scales the underlying cluster during training runs.

Answer

It forces data scientists to use only Python for feature engineering.

Answer

It eliminates the need for any data cleaning or preprocessing steps.

Question 16

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 17

You are migrating a legacy ML pipeline to Databricks. You need to ensure that the feature engineering logic used during training is identical to the logic used during real-time inference. What is the recommended approach?

Accepted Answer

Use the Databricks Feature Store to define and log features.. Consistency between training and inference (the 'training-serving skew') is a primary cause of model failure. The Databricks Feature Store acts as the single source of truth for feature definitions. By using it, you ensure that the same transformations are applied during both training and inference, eliminating discrepancies that arise from re-implementing logic in different languages or frameworks, which is critical for model reliability.

Answer

Copy the feature transformation code into both the training notebook and the inference service.

Answer

Store the features in a Parquet file and load it during both training and inference.

Answer

Implement the transformation logic as a SQL view in the Hive Metastore.

Question 18

A machine learning engineer needs to deploy a custom PyTorch model to a Databricks Model Serving endpoint. The model requires a custom pre-processing step that is not part of the standard MLflow transformers or pyfunc flavor. Which deployment approach ensures the custom logic executes reliably within the serverless serving container?

Accepted Answer

Implement a custom MLflow pyfunc PythonModel subclass that encapsulates both the PyTorch model and the custom pre-processing transformations, then log it with MLflow.. Packaging the custom pre-processing logic directly into the MLflow pyfunc model artifact by overriding the predict context ensures that all required transformations travel with the model weights. This guarantees identical execution behavior between local testing and production serverless endpoints without relying on external pipeline code.

Answer

Register the vanilla PyTorch state dict and apply pre-processing transformations inside the client application before sending HTTP payloads.

Answer

Define a separate Spark UDF inside the serving endpoint configuration file to intercept incoming JSON batches.

Answer

Store the pre-processing code in a separate volume and reference its absolute file path in the model serving endpoint environment variables.

Question 19

You are using MLflow to track experiments for a computer vision model. After several runs, you notice that the training loss is not being logged correctly because the metric name contains a space. What is the recommended way to log metrics with names that include spaces?

Accepted Answer

Replace spaces with underscores or hyphens before logging.. MLflow enforces strict naming rules for metrics and parameters, disallowing spaces. The correct approach is to replace spaces with underscores or hyphens before logging. This maintains compatibility with the MLflow tracking server, UI, and API, and avoids logging errors. Other options either do not solve the problem or introduce unnecessary complexity.

Answer

Use the `mlflow.log_metric` function with the space included; MLflow will automatically sanitize it.

Answer

Log the metric as a tag instead of a metric.

Answer

Use a custom MLflow plugin to allow spaces in metric names.

Question 20

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.