Databricks-GenAI-Assoc Assembling and Deploying Apps Practice Test 2 — 15 Questions

Question 1

A GenAI engineer has built a RAG chain with LangChain and logged it to Unity Catalog using MLflow. The chain's retriever queries a Vector Search index that is rebuilt nightly. The engineer must deploy the chain to a Databricks Model Serving endpoint so that the deployed version always reads the latest index without redeploying the endpoint. Which approach should the engineer take?

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

Accepted Answer

Log the chain with mlflow.langchain.log_model and reference the Vector Search index by name in the model signature; the endpoint resolves the index at query time.. The key requirement is that retrieval must track an index that changes nightly without redeploying the endpoint. Separating the chain from the index and referencing the index by name at query time achieves this, because each inference request performs a fresh Vector Search lookup against the current index contents. Embedding data in artifacts or caching lookups both freeze or delay freshness.

Answer

Deploy the chain as a Databricks job that runs on a schedule and writes answers to a Delta table that the application reads.

Answer

Bake the index contents into the logged model artifacts so the endpoint serves a self-contained snapshot of the embeddings.

Answer

Register the chain in Unity Catalog and configure a serving endpoint that caches index lookups for 24 hours.

Question 2

A GenAI engineer is packaging a Retrieval-Augmented Generation application as a Databricks Asset Bundle and wants `databricks bundle deploy` to also create the underlying Vector Search index and the serving endpoint that hosts the agent. Which TWO bundle resource declarations should the engineer include? (Choose two.)

Accepted Answer

A `resources.serving_endpoints` block that defines the endpoint name, config, and the served model.. Databricks Asset Bundles support declarative resource types for the components of a RAG application. The `serving_endpoints` resource creates the Model Serving endpoint that hosts the agent, and the `vector_search_indexes` resource creates the Vector Search index that powers retrieval. Together they let `databricks bundle deploy` provision both the retrieval layer and the serving layer from a single version-controlled bundle.

Answer

A `resources.model_serving` block that wraps the serving endpoint under a model-specific resource type.

Answer

A `resources.mlflow_experiments` block that registers the agent's chain as an MLflow model.

Answer

A `resources.registered_models` block that points to the Unity Catalog model to be served.

Question 3

An engineer is packaging a Mosaic AI Agent application for deployment using Databricks Asset Bundles. The bundle must (1) deploy the agent code and configuration to a target workspace from a CI pipeline, and (2) register the resulting model so a serving endpoint can reference a specific version. Which TWO bundle elements must the engineer define to accomplish this? (Choose two.)

Accepted Answer

A `targets` block mapping environments such as dev and prod to workspace hosts and configuration.. Databricks Asset Bundles express deployable resources and deployment environments through two core blocks. The resources block declares the model and serving endpoint so `bundle deploy` creates or updates them and links the endpoint to a model version. The targets block maps dev and prod to distinct workspaces and configuration, enabling CI-driven promotion. Permission, sync, and artifact blocks support deployment but do not register the model or define the endpoint.

Answer

A `permissions` block that assigns CAN_MANAGE to the CI service principal on every bundle resource.

Answer

A `sync` block that uploads the local notebook and Python files to the workspace on each deploy.

Answer

An `artifacts` block that compiles the agent into a wheel for deployment.

Question 4

A company runs a customer-facing GenAI assistant on a Mosaic AI Model Serving endpoint. During a peak-traffic incident, the endpoint began returning HTTP 429 errors and some requests timed out. The team wants the endpoint to absorb bursts without failing requests while keeping idle cost low. Which configuration change should the engineer make?

Accepted Answer

Increase the endpoint's provisioned concurrency and enable scale-to-zero when traffic drops.. HTTP 429 responses from a serving endpoint signal that the concurrency ceiling was reached. Provisioned concurrency determines how many requests can be processed simultaneously, so raising it lets the endpoint absorb traffic bursts. Pairing that with scale-to-zero keeps replicas from running when idle, controlling cost. Timeout reduction, smaller GPUs, and payload logging do not increase request-handling capacity and do not stop the throttling.

Answer

Move the endpoint to a smaller GPU SKU so more replicas fit within the same budget.

Answer

Lower the endpoint's query timeout so slow requests fail faster and free capacity.

Answer

Enable payload logging to inference tables so the endpoint can queue requests more efficiently.

Question 5

An engineer has a Mosaic AI Agent application in a Databricks Asset Bundle. The bundle defines a model serving endpoint whose configuration references a secret scope for an external LLM provider API key. The engineer runs `databricks bundle deploy` targeting the production workspace and the deployment succeeds, but at runtime the agent receives 401 Unauthorized responses from the provider. The secret scope and key already exist in the production workspace with the same name and permissions as in development. Which action should the engineer take to resolve the runtime authentication failure?

Accepted Answer

Grant the serving endpoint's service principal the READ permission on the secret scope, and verify the endpoint's environment references the secret via `{{secrets/...}}` syntax.. Model Serving resolves secrets using the endpoint's service principal identity, which needs READ on the secret scope. The 401 indicates the endpoint either lacks permission or is sending a literal placeholder instead of the resolved secret value. Granting READ and ensuring the configuration uses the `{{secrets/scope/key}}` reference syntax restores authenticated calls to the external provider.

Answer

Restart the model serving endpoint with `databricks serving-endpoints update-config` to force the secret to be re-read at startup.

Answer

Store the API key as a plaintext environment variable in the bundle's serving endpoint configuration so the endpoint can read it directly.

Answer

Add the secret scope and key names to the bundle's target configuration, then redeploy with `databricks bundle deploy`.

Question 6

An organization is deploying a GenAI application using Databricks Model Serving. Which TWO steps are required to ensure the deployment environment handles model governance and observability effectively?

Accepted Answer

Enable Unity Catalog for the registered model and its versions.. Effective model governance and observability require integrating Unity Catalog for centralized access control and MLflow for tracking inference payloads. These components allow organizations to monitor model performance, lineage, and bias over time. Mastering these tools ensures that production deployments remain compliant, transparent, and auditable, which is essential for regulated industries using generative AI to make data-driven decisions while minimizing operational risks associated with model drift and unauthorized access.

Answer

Hardcode API credentials within the model inference script.

Answer

Disable automatic scaling to maintain consistent latency.

Answer

Manually deploy the model via the standard cluster user interface.

Question 7

Which Databricks feature is primary for managing the lifecycle, versioning, and deployment readiness of custom Generative AI models?

Accepted Answer

MLflow Model Registry. MLflow Model Registry is the central feature for managing the lifecycle of models, including versioning and stage transitions. Understanding how to promote a model from 'Staging' to 'Production' is fundamental for ensuring that only tested and verified models reach the end-users. This workflow is critical for maintaining quality and stability in generative AI applications, as it provides a structured process for model evolution and deployment management.

Answer

Unity Catalog Volumes

Answer

Databricks SQL Warehouse

Answer

Delta Live Tables

Question 8

You are deploying a RAG application. You need to ensure the model uses the most recent vector data without redeploying the model. What should you use?

Accepted Answer

Implement a RAG pattern with a vector database index.. Decoupling the model logic from the data retrieval mechanism is a best practice. By using a retrieval-augmented generation pattern where the retriever fetches data dynamically from a vector store, the model remains static while the knowledge base evolves. This approach is essential for applications requiring real-time information, as it avoids the expensive and slow process of retraining or fine-tuning the base model whenever the underlying data changes.

Answer

Fine-tune the model with new data daily.

Answer

Re-register the model in MLflow with every update.

Answer

Embed all data into the model weights at deployment.

Question 9

A team has deployed a model that is experiencing high latency. How should they identify if the bottleneck is the model inference or the preprocessing code?

Accepted Answer

Use MLflow Tracing to inspect the execution pipeline.. Distinguishing between inference time and preprocessing time is vital for performance tuning. By using MLflow's tracing capabilities or custom timing logs, engineers can isolate specific segments of the request pipeline. This granular visibility allows for targeted optimization, such as optimizing the vector search or reducing prompt overhead, rather than blindly attempting to speed up the model itself, which is often significantly more resource-intensive.

Answer

Check the cluster cost in the Databricks billing console.

Answer

Re-train the model on a larger dataset.

Answer

Increase the memory limit for the inference cluster.

Question 10

Refer to the exhibit. What is the most likely cause for this error in a production RAG application?

Accepted Answer

Network connectivity between Model Serving and the vector store is misconfigured.. The error indicates a network timeout when connecting to the vector store. This is a common connectivity issue between the serving endpoint and the data source. Investigating network security groups, firewall rules, or DNS resolution issues within the Databricks environment is necessary to resolve it. Ensuring robust, low-latency connectivity to the retrieval source is fundamental for reliable RAG performance and minimizing application downtime in production.

Answer

The model is too large for the GPU memory.

Answer

The model weights are corrupted.

Answer

The input prompt is too long for the model.

Question 11

When deploying a Generative AI application, why is it recommended to use a dedicated Serving Endpoint rather than a shared interactive cluster?

Accepted Answer

Serving endpoints provide better isolation and predictable performance.. Dedicated serving endpoints provide resource isolation, performance predictability, and standardized API interfaces. Shared clusters are prone to resource contention, inconsistent environment states, and lack the operational features required for reliable production inference. Separating the serving infrastructure from development clusters is a standard architectural pattern that ensures the stability and scalability of AI-driven features in user-facing applications.

Answer

Shared clusters are cheaper to run for production.

Answer

Shared clusters cannot be used for Python-based models.

Answer

Serving endpoints automatically train the model on new data.

Question 12

An organization wants to implement 'Guardrails' on their model outputs. Which deployment strategy best facilitates this?

Accepted Answer

Deploy a wrapper model that processes the output for safety.. The best way to implement guardrails is to build a secondary validation model or use an API-based gatekeeper that intercepts and inspects model outputs before they are returned to the user. Integrating this logic into the inference pipeline within the Databricks environment ensures that all content is screened. This pattern is essential for enterprise deployments where safety, compliance, and preventing hallucinations are top priorities for generative AI applications.

Answer

Hardcode guardrails inside the base model training script.

Answer

Instruct users to manually check the output for safety.

Answer

Disable all logging to prevent guardrails from slowing down.

Question 13

Which security configuration is essential when deploying a model that accesses sensitive data stored in Unity Catalog?

Accepted Answer

Use a Service Principal with scoped permissions for the model.. Accessing sensitive data requires strict adherence to the principle of least privilege. Assigning a specific Service Principal to the model serving endpoint ensures that it only has access to the data necessary for inference. This configuration is critical for security, as it limits the potential blast radius if the model endpoint is ever compromised, maintaining data privacy in line with organizational governance policies.

Answer

Grant the endpoint access to the entire Unity Catalog.

Answer

Store credentials in plaintext in the model code.

Answer

Bypass Unity Catalog to use local file system access.

Question 14

Refer to the exhibit. What is the impact of min_instances: 0 on this deployment?

Accepted Answer

The endpoint will scale to zero when idle, reducing costs.. Setting min_instances to 0 in Databricks Model Serving enables 'scale-to-zero'. This feature is highly effective for cost optimization, as it automatically shuts down the GPU resources when no traffic is present. Upon receiving a new request, the endpoint will automatically scale up, though this may introduce a slight 'cold start' latency. This trade-off is often acceptable for non-critical or low-traffic services to drastically reduce infrastructure costs.

Answer

The endpoint will be permanently disabled.

Answer

The endpoint will have higher latency than a fixed instance.

Answer

The endpoint will be restricted to CPU usage only.

Question 15

Which CI/CD approach for model deployment best minimizes downtime during a model update?

Accepted Answer

Use a blue-green deployment pattern.. A blue-green deployment strategy is the gold standard for minimizing downtime. By spinning up a new version of the model (green) alongside the existing one (blue) and switching traffic only after verifying the health of the green endpoint, engineers ensure zero downtime and an immediate rollback path if issues occur. This approach is critical for high-availability AI services where any interruption would negatively impact users.

Answer

Delete the old model and immediately deploy the new version.

Answer

Deploy all updates directly to the production endpoint.

Answer

Schedule deployments during peak traffic hours.