Databricks · DCGAE
Databricks Certified Generative AI Engineer Associate Practice Tests
Validates the ability to design, develop, and deploy LLM-powered solutions on Databricks, covering RAG application design and data preparation, prompt engineering and retrieval chains, model serving and deployment, evaluation and monitoring for quality and safety, and governance with Unity Catalog.
Exam Details
Practice Questions
620
≈ 13 practice exams
Duration
90 minutes
Passing Score
70%
Difficulty
AssociateLast Updated
Sep 2026
Databricks Certified Generative AI Engineer Associate (DCGAE) Practice Exam
Application Development is the heaviest domain at 30%, followed by Assembling and Deploying Applications at 22%, Design Applications at 14%, Data Preparation at 14%, Evaluation and Monitoring at 12%, and Governance at 8%. This 620-question practice bank is built to match that split, so LLM chain construction, prompt augmentation, and Model Serving deployment scenarios get real depth instead of a token appearance. Watch the freshness of whatever you study from: Databricks refreshed the exam guide on March 18, 2026, and the live exam now covers Agent Bricks, MCP server integration, AI Gateway, Genie Spaces, and ai_query() batch inference alongside the classic RAG material.
Test day means 45 scored multiple-choice and multiple-select questions in 90 minutes, delivered as an online proctored exam through Webassessor. Databricks may mix in unscored pilot questions that are not identified and do not affect your result, with extra time already factored in to cover them. There is no published passing percentage; you receive a pass or fail outcome with a per-domain breakdown, and candidates commonly target roughly 70% to be safe. All ML code on the exam is written in Python, SQL can appear for non-ML data manipulation, and the exam is offered in English, Japanese, Korean, and Brazilian Portuguese.
There are no formal prerequisites, but Databricks recommends six months of hands-on generative AI work plus the Generative AI Engineering with Databricks courses in Databricks Academy. Registration costs $200 USD per attempt, and the credential expires after two years; recertifying means retaking the full exam that is live at that time, which matters here because the content shifts fast. Start with the 30 free questions, then work through the full 620-question bank until your accuracy holds steady across all six domains.
Exam Domain Breakdown
Application Development30%
Assembling and Deploying Applications22%
Design Applications14%
Data Preparation14%
Evaluation and Monitoring12%
Governance8%
Exam Overview
The Databricks Certified Generative AI Engineer Associate certification validates an individual's ability to design, develop, and deploy large language model (LLM)-powered solutions on the Databricks platform. The exam tests practical competency across the full generative AI engineering lifecycle, including decomposing complex requirements into multi-stage reasoning pipelines, selecting appropriate models from both open-source and proprietary ecosystems, and implementing retrieval-augmented generation (RAG) applications using Databricks-native tooling.
Certified professionals are expected to demonstrate hands-on proficiency with key Databricks technologies: Vector Search for semantic similarity and document retrieval, Model Serving for scalable endpoint deployment, MLflow for experiment tracking and lifecycle management, and Unity Catalog for data governance and access control. All machine learning code on the exam is in Python; SQL may appear for non-ML data manipulation tasks. The certification is valid for two years, after which recertification requires retaking the current version of the exam.
Who Should Take This Exam
This certification is designed for practitioners actively building and deploying AI systems in enterprise environments, including AI Engineers, Generative AI Engineers, LLM Engineers, AI Solution Architects, MLOps Engineers, and Data Scientists with a focus on LLM or RAG workflows. It is particularly well-suited for engineers who work within the Databricks ecosystem and need to demonstrate production-level competency in generative AI solution development.
Candidates should have at least six months of hands-on experience developing generative AI solutions, practical familiarity with Python-based ML pipelines, and working knowledge of frameworks such as LangChain or LangGraph. Experience with Databricks-specific tools—MLflow, Unity Catalog, Vector Search, and Model Serving—is strongly recommended before attempting the exam.
Prerequisites
There are no formal prerequisites required to register for this exam; any candidate may attempt it. However, Databricks strongly recommends at least six months of hands-on experience in generative AI solution development before sitting for the certification.
Recommended technical knowledge includes Python proficiency (especially for model pipelines and application orchestration), familiarity with LLM concepts such as context windows, tokenization, and prompt engineering techniques (zero-shot, few-shot, chain-of-thought), and practical experience with LangChain or similar orchestration frameworks. Candidates should also be comfortable using Databricks-native tools including MLflow for experiment tracking, Unity Catalog for governance, Vector Search for embedding-based retrieval, and Model Serving for endpoint deployment.
Exam Format
The exam consists of approximately 45 scored multiple-choice and multiple-select questions to be completed within 90 minutes. It is delivered as a proctored online exam, meaning candidates complete it remotely under live or automated proctoring; no external aids are permitted. The exam is available in English, Japanese, Brazilian Portuguese, and Korean. The registration fee is $200 USD (local taxes may apply).
The passing score is 70%. Databricks notes that exams may include additional unscored items used to gather statistical data for future exam development; these items are not identified and do not affect the final score, meaning the total number of questions delivered may be slightly higher than the 45 scored items. The certification remains valid for two years, after which candidates must retake the current exam version to recertify.
Skills Measured
- 1.Application Development (30%): Implementing LLM chains and RAG pipelines using LangChain and LangGraph; applying context injection, few-shot prompting, and agentic AI patterns; building Python-based orchestration pipelines; integrating with external APIs and databases for dynamic data retrieval.
- 2.Assembling and Deploying Applications (22%): Deploying solutions to Databricks Model Serving endpoints; packaging models using MLflow pyfunc; managing model lifecycle with MLflow Tracking and Model Registry; optimizing Vector Search endpoints; incorporating Foundation Model APIs; configuring inference tables for logging.
- 3.Design Applications (14%): Designing multi-stage reasoning pipelines; selecting between open-source and proprietary LLMs based on requirements; applying prompt engineering techniques including zero-shot, few-shot, prompt chaining, and meta-prompts; defining agent tools and safety optimization strategies; understanding system and user prompt roles.
- 4.Data Preparation (14%): Processing and ingesting documents for RAG pipelines; selecting chunking strategies and embedding models; storing and indexing vectors; applying filtering techniques; leveraging Delta Lake for structured data storage within generative AI workflows.
- 5.Evaluation and Monitoring (12%): Evaluating retriever and LLM performance using MLflow; implementing LLM-as-a-Judge approaches for automated scoring; tracking metrics such as context precision, toxicity, and perplexity; configuring continuous logging and inference request analysis for production monitoring.
- 6.Governance (8%): Applying Unity Catalog for data access control, auditing, and compliance; implementing ethical AI frameworks and guardrails; managing PII exposure risks and security controls; ensuring regulatory compliance for enterprise AI deployments.
Study Tips
- Complete the official Databricks Academy self-paced course 'Generative AI Engineering with Databricks' before attempting the exam — it maps directly to all six exam domains and includes hands-on notebooks using Databricks Community Edition.
- Build at least one end-to-end RAG application from scratch using Databricks Vector Search, MLflow, and Model Serving. The exam heavily emphasizes Application Development (30%) and Assembling & Deploying Apps (22%), so practical lab experience with these tools is essential.
- Study MLflow's evaluation capabilities in depth, including how to configure LLM-as-a-Judge scoring and batch evaluation runs. The Evaluation and Monitoring domain tests specific API usage and metric interpretation, not just conceptual knowledge.
- Review the Unity Catalog documentation focused on AI governance use cases — access control for model endpoints, audit logging, and securing inference pipelines. The Governance domain (8%) is often under-studied but has clearly defined testable content.
- Practice under timed conditions using third-party practice exams on platforms like Udemy or Whizlabs. With 45 questions in 90 minutes, you have about two minutes per question; scenario-based questions require quick identification of the correct Databricks tool or API for a given use case.
- Study the LangChain and LangGraph integration patterns specific to Databricks, including how chains are structured, how retrievers are connected to vector stores, and how agents are defined with tools. All ML code on the exam is Python, so be comfortable reading and interpreting LangChain pipeline code.
- Review Databricks Foundation Model APIs and the difference between external model endpoints and Databricks-hosted models. Understanding when to use pay-per-token Foundation Model APIs versus deploying custom models to serving endpoints is a recurring theme in the deployment and design domains.
Career Benefits
Professionals holding this certification are positioned for roles at the intersection of software engineering and applied AI, including Generative AI Engineer, LLM Engineer, AI Solution Architect, and MLOps Engineer. Generative AI engineering roles command some of the highest compensation in the technology sector, with average salaries reported around $214,000 annually in the United States; the certification directly signals enterprise-grade deployment skills that go beyond prototyping or research experience.
The generative AI applications market is projected to grow at a CAGR exceeding 46% through 2030, and employer demand for engineers who can bridge the gap between experimental LLM work and production-ready Databricks deployments continues to outpace supply. As Databricks is widely adopted across Fortune 500 companies for data and AI workloads, this certification carries strong recognition among employers already invested in the Databricks ecosystem. It complements other Databricks credentials (such as the Data Engineer Associate or ML Professional certifications) for practitioners building a comprehensive Databricks certification portfolio.
Sample Questions
5 sample questions with answers and explanations. The full bank has 620 questions, enough for 13 full-length practice exams.
Preview — answers shown1. A Generative AI Engineer is building a RAG application over a product inventory dataset stored as a CSV file. Each row represents a single product with columns for product ID, name, category, price, and description. What is the recommended chunking strategy for this structured dataset? (Select one!)
AApply recursive character splitting with a 512-token chunk size and 10% overlap so the model can infer relationships between adjacent product rows
BApply fixed-size chunking of 256 tokens and truncate any row that exceeds the limit to maintain uniform chunk sizes across the dataset
CTreat each row as one chunk, with each product row becoming a single independent document for indexing
DUse semantic chunking to group products with similar descriptions into topic-coherent segments, maximizing embedding cohesion across the catalog
Explanation
For structured data stored in CSV format, the correct approach is one row per chunk. Each product is a self-contained, atomic unit of information, and treating each row as its own document ensures the embedding captures the full product context and that retrieval returns complete, actionable records. Recursive character splitting applies character-boundary logic designed for unstructured prose; it will arbitrarily break mid-row or merge rows from different products, destroying the row-level semantics of structured data. Semantic chunking clusters by topic similarity across continuous text, which is inappropriate for tabular data where each row's identity must remain intact and independent. Fixed-size chunking with truncation permanently discards field values on longer rows, introducing data loss with no retrieval quality benefit.
2. A Generative AI Engineer at Viewpoint Studios is building a content policy enforcement system for user-uploaded media. Each submission may include a thumbnail image, a short video clip, or both, combined with a text description. The system must analyze all provided modalities in a single model call to detect policy violations without requiring an additional preprocessing pipeline. Which Foundation Model API model should the engineer select? (Select one!)
ABGE-large for its high-dimensional embedding representations of multimodal content
BGemini 2.5 Flash for its native support of text, image, video, and audio in a single model call
CLlama 4 Maverick for its large context window and support for text and image inputs
DClaude Sonnet 4 for its strong instruction-following capability across text and image inputs
Explanation
Gemini 2.5 Flash natively supports all four input modalities — text, image, video, and audio — within a single model call through the Foundation Model API. This native video support is the decisive differentiator that allows thumbnail images and video clips to be processed simultaneously without any additional preprocessing pipeline. Claude Sonnet 4 supports text and image inputs within the Foundation Model API but does not support video as an input modality, making it incapable of natively ingesting video clips for policy analysis. Llama 4 Maverick similarly supports text and image inputs only and cannot process video content without a separate frame-extraction step that routes the work outside the model's native capabilities. BGE-large is an embedding model that produces vector representations for retrieval and similarity search; it cannot interpret visual or video content, cannot detect policy violations, and cannot generate structured text output of any kind.
3. A Generative AI Engineer's upstream embedding pipeline normalizes all document vectors to unit length before writing them to a Delta table. The team needs to confirm that Databricks AI Search can support cosine similarity semantics for a document deduplication system. Which statement accurately describes how cosine similarity is achieved in Databricks AI Search? (Select one!)
APass distance_metric='cosine' in the index creation parameters to enable native cosine distance calculation
BSwitch to a Direct Access Index, which exposes an alternative cosine distance mode unavailable in Delta Sync Indexes
CCosine similarity is not supported; Databricks AI Search only exposes L2 (Euclidean) distance with no workaround
DDatabricks AI Search uses L2 distance internally, but pre-normalizing embeddings to unit length before indexing produces results mathematically equivalent to cosine similarity
Explanation
Databricks AI Search uses the Hierarchical Navigable Small World algorithm with L2 (Euclidean) distance as its internal metric. For unit-normalized vectors, L2 distance and cosine distance are mathematically equivalent and produce identical rankings because the magnitude of all vectors is 1. Since the engineer's pipeline already normalizes embeddings to unit length before storage, the L2-based index inherently delivers cosine similarity behavior with no additional configuration required. There is no distance_metric parameter in the AI Search index creation API. Cosine similarity is effectively supported through normalization rather than a configurable metric switch. The index type (Delta Sync versus Direct Access) does not affect the underlying distance metric used by the HNSW algorithm.
4. A Generative AI Engineer at Pembury Systems configured an inference table on a production Model Serving endpoint to capture all requests and responses for compliance auditing. A data engineer later modified the inference table schema to add two columns required by a downstream analytics pipeline. The compliance team subsequently discovered a logging gap covering the three weeks since the schema was changed. What caused this gap? (Select one!)
AThe two new columns increased average row size beyond the 10 MiB per-payload limit, causing all log entries after the change to be silently dropped.
BInference tables stop logging when the table schema is changed because the write pipeline requires the schema to exactly match the fixed set of columns Databricks defines, and any modification breaks the write path indefinitely.
CInference tables enforce a 21-day rolling retention window, and schema changes trigger an automatic purge of all data within that window.
DInference tables automatically suspend logging for up to 30 days when schema changes are detected to allow internal write buffers to rebuild.
Explanation
Databricks inference tables stop logging, or become corrupted, whenever the table schema is changed after the serving endpoint is configured, when the table is renamed, or when the table is deleted. The logging pipeline depends on a fixed schema precisely matching the columns Databricks writes: databricks_request_id, client_request_id, date, timestamp_ms, status_code, sampling_fraction, execution_time_ms, request, response, and request_metadata. Altering that schema breaks the internal write path and logging stops indefinitely. The correct approach is to leave the inference table schema untouched and apply downstream transformations or column additions in a separate ETL process. There is no automatic suspension period or write-buffer rebuild behavior triggered by schema changes. The 10 MiB payload limit causes individual oversized payloads to be dropped, not all subsequent logging to cease. Inference tables do not implement a rolling retention window or automatically purge data following schema modifications.
5. A Generative AI Engineer at Kestrel Analytics has been running a text summarization service on Databricks Foundation Model APIs using pay-per-token pricing for six months. The service now processes 50,000 summarization requests per day with a predictable, consistent traffic pattern during business hours. The platform team wants to reduce per-request latency variance and lower costs at this volume. Which change should the engineer make? (Select one!)
AEnable inference tables to profile latency distributions before committing to any infrastructure change
BReduce maximum prompt length to decrease per-call token consumption and shared-capacity contention
CMigrate to provisioned throughput by specifying appropriate minimum and maximum token allocations
DRoute requests through an external model via AI Gateway to avoid shared-capacity queuing
Explanation
Provisioned throughput is the recommended configuration for production workloads with consistent, high-volume traffic. It allocates dedicated model capacity that eliminates the latency variance inherent in shared pay-per-token infrastructure and typically delivers a lower effective cost per token at sustained scale compared to on-demand pricing. The min_provisioned_throughput and max_provisioned_throughput parameters allow the engineer to right-size the allocation to the known traffic pattern. Pay-per-token is intended as the starting point for development and early-stage exploration, not for high-volume predictable production workloads. Routing through an external model via AI Gateway adds network hops and operational complexity without providing the dedicated capacity allocation that resolves latency variance at its root. Enabling inference tables adds monitoring visibility into request patterns but does not modify the underlying serving configuration or address the infrastructure source of latency variance. Truncating prompts reduces per-call token counts but degrades summarization quality and does not solve latency variance, which originates from shared infrastructure contention rather than from prompt length.
Frequently Asked Questions
How many questions are on the Databricks Generative AI Engineer Associate exam?
45 scored multiple-choice and multiple-select questions in 90 minutes. Databricks may add unscored pilot questions that are not identified and do not count toward your result; extra time is already built in for them.
What is the passing score for the Databricks Generative AI Engineer Associate exam?
Databricks does not publish a fixed passing percentage. You receive a pass or fail result along with a per-domain score breakdown; candidates commonly aim for roughly 70% correct to pass comfortably.
How much does the Databricks Generative AI Engineer Associate exam cost?
$200 USD per attempt, plus any applicable local taxes. Registration is handled through the Webassessor platform and the exam is delivered online proctored.
What domains does the Databricks Generative AI Engineer Associate exam cover?
Six domains: Application Development (30%), Assembling and Deploying Applications (22%), Design Applications (14%), Data Preparation (14%), Evaluation and Monitoring (12%), and Governance (8%).
Are there prerequisites for the Databricks Generative AI Engineer Associate exam?
No prerequisites are required. Databricks recommends six or more months of hands-on generative AI solution work and the Generative AI Engineering with Databricks self-paced courses in Databricks Academy before attempting it.
How long is the Databricks Generative AI Engineer Associate certification valid?
Two years. To recertify you must retake the full exam version that is live at that time; there is no shorter renewal exam or continuing-education path.
Has the Databricks Generative AI Engineer Associate exam changed recently?
Yes. The exam guide was refreshed on March 18, 2026 (after an interim February 2026 update) and now tests Agent Bricks, managed and custom MCP servers, AI Gateway, Genie Spaces, custom Scorers, and prompt lifecycle management in addition to the original RAG-focused content.
More Databricks Practice Exams
Databricks Certified Data Analyst Associate
DCDAA · 627 questions
Databricks Certified Data Engineer Associate
DCDEA · 628 questions
Databricks Certified Data Engineer Professional
DCDEP · 628 questions
Databricks Certified Machine Learning Associate
DCMLEA · 630 questions
Databricks Certified Machine Learning Professional
DCMLEP · 622 questions
Databricks Certified Associate Developer for Apache Spark
DCASD · 604 questions
$17.99
One-time access to this exam
620 questions (13 practice exams' worth)
Unlimited timed exam simulations
Yours forever — no subscription
Detailed explanations
Free preview stays available