Free Databricks Exam Questions

Core Databricks Generative AI Technologies:

• Vector Search (Mosaic AI Vector Search) - Semantic similarity searches, creating and querying vector search indexes
• Model Serving - Deploying models and solutions, controlling access to resources from model serving endpoints
• MLflow - Managing solution lifecycle, registering models to Unity Catalog, evaluating model performance in RAG applications
• Unity Catalog - Data governance for generative AI applications, managing RAG application data
• RAG (Retrieval Augmented Generation) - Building RAG applications, chunking strategies, document extraction and processing
• LLM Chains - Building and deploying LLM chains, prompt engineering, prompt generation and evaluation
• Foundation Model APIs - Serving LLM applications using Foundation Model APIs
• Agent Framework - Developing agentic systems, creating tools for multi-stage reasoning
• LangChain and Similar Tools - Using LangChain and similar frameworks for GenAI application development
• Inference Logging and Agent Monitoring - Tracking live LLM endpoints, assessing deployed RAG application performance
• Delta Lake - Writing chunked text into Delta Lake tables in Unity Catalog
• Guardrails and Safety - Implementing LLM guardrails, masking techniques, protecting against malicious inputs

Exam Sections (6 Main Domains):

1. Design Applications - Designing prompts, selecting model tasks, choosing chain components, translating business use cases into AI pipeline inputs/outputs, defining and ordering tools for multi-stage reasoning
2. Data Preparation - Applying chunking strategies, filtering extraneous content, choosing Python packages for document extraction, writing chunked text to Delta Lake, identifying needed source documents, evaluating retrieval performance, designing retrieval systems with advanced chunking, understanding re-ranking in information retrieval
3. Application Development - Creating tools for data retrieval, selecting LangChain/similar tools, understanding prompt format impacts, qualitatively assessing responses, selecting chunking strategies, augmenting prompts with context, adjusting LLM responses, implementing LLM guardrails, writing metaprompts, building agent prompt templates, selecting best LLM and embedding models, selecting models from hubs/marketplaces, utilizing Agent Framework
4. Assembling and Deploying Applications - Coding chains using pyfunc models, controlling access to model serving endpoints, coding simple chains, creating RAG applications, registering models to Unity Catalog, deploying endpoints, creating and querying Vector Search indexes, serving LLM applications with Foundation Model APIs, understanding Mosaic AI Vector Search, applying ai_query() for batch inference
5. Governance - Using masking techniques as guardrails, selecting guardrail techniques, recommending alternatives for problematic text mitigation, using legal/licensing requirements for data sources
6. Evaluation and Monitoring - Selecting LLM choices based on evaluation metrics, selecting key metrics to monitor, evaluating model performance using MLflow, using inference logging, controlling LLM costs, using inference tables and Agent Monitoring, identifying evaluation judges requiring ground truth, comparing evaluation and monitoring phases

Foundation Skills Tested:

• Designing and implementing LLM-enabled solutions using Databricks
• Problem decomposition to break down complex requirements into manageable tasks
• Choosing appropriate models, tools, and approaches from the generative AI landscape
• Building and deploying performant RAG applications
• Creating and managing LLM chains that take full advantage of Databricks
• Using Vector Search for semantic similarity searches
• Deploying models and solutions using Model Serving
• Managing solution lifecycle with MLflow
• Implementing data governance with Unity Catalog for GenAI applications
• Prompt engineering, prompt generation, and evaluation
• Implementing guardrails and safety measures for GenAI applications
• Evaluating and monitoring deployed GenAI applications