- What the Credential Actually Is After the October 2026 Rename
- The Three Domains and What Each One Tests
- Domain 1: Design and Develop Database Solutions
- Domain 2: Secure, Optimize, and Deploy Database Solutions
- Domain 3: Implement AI and Analytics Capabilities
- Passing Score, Renewal, and What We Cannot Confirm
- Who Should Pursue This Certification
- Sequencing Your Preparation by Domain
- Related Resources
- Frequently Asked Questions
- DP-420 is now Microsoft Certified: Azure Cosmos DB AI Developer Associate, renamed effective October 6, 2026; the exam code is unchanged.
- The exam is titled Building AI Applications with Azure Cosmos DB and is organized into three domains, not the old five.
- Domain 1 (Design and develop database solutions) carries the largest weight at 40-45%, so data modeling and partition keys deserve the most study time.
- The passing score is 700, and renewal is annual through Microsoft's free online assessment.
What the Credential Actually Is After the October 2026 Rename
If you have seen older study material calling this credential the "Azure Cosmos DB Developer Specialty," you are looking at the previous name. Microsoft renamed the certification to Microsoft Certified: Azure Cosmos DB AI Developer Associate effective October 6, 2026. The exam you sit for is still DP-420, and it is now titled Building AI Applications with Azure Cosmos DB.
This is a rename and a refocus, not a brand-new certificate and not a replacement exam. The same assigned exam identity carries forward, but the skills outline has been reorganized around three domains and shifted toward AI-enabled application development. Candidates who studied from a pre-rename outline will find familiar territory in core Cosmos DB engineering, plus considerably more emphasis on vector search, agent memory, and analytics integration.
The Three Domains and What Each One Tests
The current skills measured group the exam into three areas. The weights below are the ranges published for this exam; plan your time around them rather than around any rumored point values.
| Domain | Weight | Core Theme |
|---|---|---|
| Design and develop database solutions | 40-45% | Data modeling, partition keys, SDK operations, queries, change feed, AI-assisted development |
| Secure, optimize, and deploy database solutions | 30-35% | Security controls, backup and failover, indexing and performance, monitoring, Fleet |
| Implement AI and Analytics capabilities in database solutions | 20-25% | Vector, full-text, and hybrid search, RAG, agent memory, Fabric integration, Spark |
For a deeper walk through how these areas relate to one another, see the complete guide to all three DP-420 content areas. What follows is a candidate-oriented summary of what each domain demands in practice.
Domain 1: Design and Develop Database Solutions
At 40-45%, this is the heaviest domain, and it is where most of the exam's scenario-based judgment lives. It is also where developers who only use Cosmos DB casually tend to have gaps, because correct answers hinge on understanding request units and partitioning rather than just calling an API.
Resource models, SDKs, and account settings
You need to know how accounts, databases, containers, and items relate, and how SDK clients are configured and used.
- Consistency levels and the trade-offs each one makes between latency, availability, and read guarantees
- Request units and provisioned versus other throughput options at the container and database level
- Account-level and container-level settings that affect behavior and cost
Working with data: queries and operations
Expect to evaluate code and choose the most efficient way to read or modify data.
- Query design, paging, and continuation tokens
- Point reads versus queries, and why a point read is usually the cheaper choice when you know the id and partition key
- CRUD operations, patch operations, bulk execution, batch operations, and transactions
- Time to live (TTL) and optimistic concurrency control
Data modeling and partition keys
This is the intellectual center of Cosmos DB, and the topic where weak candidates lose the most points.
- Choosing partition keys that distribute storage and request load evenly
- Hierarchical partition keys, synthetic partition keys, and multitenant patterns
- Recognizing anti-patterns such as hot partitions and cross-partition fan-out
- Embedding versus referencing related data
Change feed and AI-assisted development
The change feed remains a core integration mechanism, and the rename adds explicit AI-assisted development content.
- Change feed modes, lease handling, and copy jobs
- AI-assisted development workflows, including Agent Kit
Key Takeaway
When a question describes uneven load or throttling, think partition key before you think throughput. Raising provisioned throughput is rarely the right answer when the real problem is a skewed partitioning strategy.
Domain 2: Secure, Optimize, and Deploy Database Solutions
This domain, at 30-35%, covers what it takes to run Cosmos DB reliably in production. It rewards candidates with operational experience, and it is the area where reading documentation without hands-on practice leaves visible gaps.
Security
Questions here test authentication and authorization choices, network controls, and dynamic data masking. You should be able to reason about when to use identity-based access versus keys, how to restrict network exposure, and how masking protects sensitive fields from certain readers.
Resilience: backups, restores, and failover
Know how backup and restore work, how multiregion configurations behave, and how partition failover fits into availability planning. Scenario questions typically ask you to match a recovery objective to the right configuration.
Performance and monitoring
Expect to analyze query and indexing performance, interpret metrics and latency, and use Azure Monitor and Log Analytics to diagnose problems. Understanding how indexing policy choices affect both RU cost and write overhead is a recurring theme.
Fleet
The current outline adds Fleet-related material: managing throughput across accounts and running analytics across them. Treat it as its own study item rather than assuming general throughput knowledge will carry you through.
Domain 3: Implement AI and Analytics Capabilities
Although it carries the smallest weight at 20-25%, this domain is what distinguishes the renamed credential from its predecessor, and it is the area most likely to be unfamiliar even to experienced Cosmos DB engineers.
Search: full-text, vector, and hybrid
You must understand how Cosmos DB supports retrieval for AI workloads.
- Full-text search, vector search, and hybrid search, and when each is appropriate
- Vector data types, vector indexes, size considerations, and partitioning implications
- Sharded DiskANN and its role in scaling vector search
- Performance tuning for vector queries and building retrieval-augmented generation (RAG) solutions
Agent memory
The exam covers using Cosmos DB as the memory layer for AI agents.
- Durable memory that persists across sessions
- Conversation-state memory for ongoing interactions
- Semantic memory that retrieves relevant context by meaning
Analytics and Fabric integration
Candidates should be able to distinguish the operational database from its analytical counterpart.
- Cosmos DB versus Cosmos DB in Fabric, and the use case for each
- Mirroring operational data into Fabric
- Querying with T-SQL and Spark
- Reading and writing data through the Spark connector
Passing Score, Renewal, and What We Cannot Confirm
The passing score for DP-420 is 700, on Microsoft's standard scaled scoring. A scaled score is not a straight percentage of questions answered correctly, so avoid trying to reverse-engineer a raw-score target. For more detail, see our page on the DP-420 passing score.
Microsoft role-based certifications at the associate level require renewal, and for this credential renewal is annual and is done through a free online assessment on Microsoft Learn rather than by retaking the full exam.
Who Should Pursue This Certification
The Associate level signals working competence rather than entry-level familiarity. The people who get the most from it are those who build or operate applications on Cosmos DB, and increasingly those building AI-enabled applications that need a database for retrieval and agent state.
- Application developers building services on Azure that store and query JSON data at scale
- AI and ML engineers who need a vector-capable store for RAG pipelines and agent memory
- Cloud and data engineers responsible for the performance, security, and cost of Cosmos DB deployments
- Solution architects who need to justify partitioning and consistency decisions
Employers searching for Cosmos DB skills typically list them alongside broader Azure, API, and data platform requirements, so the credential works best as supporting evidence next to project experience. For a realistic look at roles, see our notes on DP-420 jobs, and for the economics, the ROI analysis. If you are weighing eligibility, the requirements page addresses prerequisites.
Sequencing Your Preparation by Domain
Rather than a generic study calendar, order your preparation around how the domains depend on each other. Partitioning knowledge from Domain 1 underpins the performance questions in Domain 2 and the vector partitioning questions in Domain 3, so it should come first.
Domain 1 foundations
- Build containers with deliberately good and bad partition keys and observe RU behavior
- Practice point reads, patch, bulk, and transactional batch in an SDK
- Wire up a change feed processor and examine leases
Domain 2 operations
- Configure authentication, network restrictions, and masking
- Test backup and restore, then review multiregion and failover options
- Read query metrics, tune indexing policy, and explore Azure Monitor and Log Analytics
Domain 3 AI and analytics
- Create a vector index and compare vector, full-text, and hybrid queries
- Prototype an agent memory design covering durable, conversation, and semantic memory
- Set up mirroring to Fabric and run T-SQL and Spark queries
Integration and practice
- Work through mixed-domain scenarios on the DP-420 practice tests
- Revisit weak areas, weighting Domain 1 revision most heavily
Adjust the length to your background. Someone already running Cosmos DB in production may compress the first weeks, while someone coming from relational databases should expect to spend longer on the document and partitioning mindset. Our full study guide and the one-page cheat sheet are useful companions, and the training overview lists learning formats.
Key Takeaway
Build things. The exam is scenario-driven, and candidates who have actually watched a hot partition throttle or a vector query degrade recognize the pattern in a question far faster than those who only read about it. Use the practice test site to check whether you can apply what you built.
Related Resources
If you are still orienting yourself, these pages answer the most common foundational questions: what DP-420 is, what DP-420 stands for, and how the credential relates to earnings. For hands-on question practice, return to the main practice test site.
Frequently Asked Questions
No. The exam is still DP-420. Microsoft renamed the certification effective October 6, 2026, and retitled the exam Building AI Applications with Azure Cosmos DB. It is the same assigned exam identity, not a new certificate or a replacement.
Three: Design and develop database solutions (40-45%), Secure, optimize, and deploy database solutions (30-35%), and Implement AI and Analytics capabilities in database solutions (20-25%). Older five-domain outlines are outdated.
The passing score is 700 on Microsoft's scaled scoring system. Because scores are scaled, 700 does not equal a fixed percentage of questions answered correctly.
Renew annually by completing Microsoft's free online assessment on Microsoft Learn. You do not need to retake the full exam to renew.
These details are not established by Microsoft's study-guide page, and fees vary by region. Check Microsoft's official exam page and the Pearson VUE booking flow for the current figures before you schedule.