Training: Build AI applications with Azure Cosmos DB (DP-420)

Ref. DP-420T00

Download PDF

Duration:

days

Exam:

Optional

Level:

Intermediate

Funding:

Eligible

Home > Trainings > IT Pro > Cloud Computing > Build AI applications with Azure Cosmos DB (DP-420)

Build AI applications with Azure Cosmos DB Training (DP-420)

You need to build globally distributed applications with low-latency data access but find traditional relational databases limiting? Azure Cosmos DB is Microsoft’s flagship NoSQL database service, designed for mission-critical applications requiring instant response times worldwide. Without proper design patterns, Cosmos DB projects can quickly become costly and underperforming.

This official Microsoft DP-420 training gives you, in 4 days, the skills to build AI applications with Azure Cosmos DB. Guided by MCT-certified trainers, you learn to model and partition NoSQL data, configure throughput and consistency, and optimize cost and performance, then move on to what matters most today: vector and hybrid search, RAG applications and AI agent memory. This training prepares you for the Azure Cosmos DB AI Developer Associate certification. Sessions are available in person in Geneva and Lausanne or via virtual classroom.

Participant Profiles

Developers building AI applications and agents on Azure
Cloud-native application developers using Azure Cosmos DB
AI engineers implementing vector search and RAG
Solution architects and data engineers designing storage for AI applications

Objectives

Model and partition Azure Cosmos DB for NoSQL data around your access patterns
Develop with the Azure Cosmos DB SDK: operations, queries, transactions and change feed
Implement vector, full-text and hybrid search
Build RAG applications and memory stores for AI agents
Secure, back up and distribute a solution across multiple regions
Optimize indexing, monitor performance and control request unit consumption

Prerequisites

Knowledge of Microsoft Azure and ability to navigate the Azure portal (AZ-900 equivalent)
Experience writing in an Azure-supported language at the intermediate level. (C#, JavaScript, Python, or Java)
Ability to write code to connect and perform operations on a SQL or NoSQL database product. (SQL Server, Oracle, MongoDB, Cassandra or similar)

Course Content

Module 1: Explore Azure Cosmos DB for NoSQL

What is Azure Cosmos DB for NoSQL
How Azure Cosmos DB for NoSQL works
Evaluate Azure Cosmos DB for your workload
Explore resources and basic operations

Module 2: Configure resources, throughput, and consistency

Evaluate throughput and request unit requirements
Choose a throughput model
Configure database- and container-level throughput
Evaluate and configure consistency levels
Configure time to live for data retention

Module 3: Connect to Azure Cosmos DB with the SDK

Understand and import the SDK
Connect using a client singleton
Configure connectivity mode and settings
Enable offline development with the emulator
Handle connection errors, logging, and parallelism

Module 4: Implement Azure Cosmos DB operations with the SDK

Perform point read operations
Create, replace, upsert, and delete items
Update items with partial document patch
Configure item time to live
Implement optimistic concurrency control
Execute transactional batch operations
Process bulk operations

Module 5: Query data in Azure Cosmos DB for NoSQL

Understand the SQL query language
Create and execute queries
Implement correlated subqueries
Create cross-product queries
Page results with continuation tokens

Module 6: Design a data modeling and partitioning strategy

Identify aggregate boundaries and access patterns
Map aggregates to items: embed or reference
Denormalize and pre-aggregate for performance
Select a partition key
Implement hierarchical partition keys
Implement synthetic keys and multitenant partitioning
Identify anti-patterns and estimate cost

Module 7: Process the Azure Cosmos DB for NoSQL change feed

Understand the change feed
Implement a change feed consumer
Choose a change feed mode
Handle events with Azure Functions
Move data with container copy jobs

Module 8: Implement AI-assisted development tools for Azure Cosmos DB

Understand the AI-assisted Azure Cosmos DB toolset
Use the Azure Cosmos DB MCP server
Use the agent toolkits
Choose the right tool and understand its scope

Module 9: Secure an Azure Cosmos DB solution

Implement Microsoft Entra authentication
Configure key-based authentication and rotation
Implement role-based access control
Configure network security
Configure dynamic data masking

Module 10: Configure backup and restore in Azure Cosmos DB

Evaluate backup options
Configure continuous backup
Perform point-in-time recovery
Describe Azure backup for Azure Cosmos DB

Module 11: Design multi-region availability and failover in Azure Cosmos DB

Explore global distribution and availability
Add regions and route reads
Configure multi-region writes
Resolve write conflicts
Configure account failover
Configure partition-level failover

Module 12: Design and optimize an indexing strategy in Azure Cosmos DB

Understand indexing policies
Configure indexing policies for workloads
Configure composite and spatial indexes
Configure computed properties
Implement global secondary indexes

Module 13: Analyze and tune query and operation performance in Azure Cosmos DB

Evaluate request unit consumption and high-cost queries
Analyze query performance metrics
Analyze index performance metrics
Optimize cross-partition queries

Module 14: Monitor and troubleshoot Azure Cosmos DB

Interpret response codes and diagnose throttling
Monitor latency and partition utilization
Configure Azure Monitor alerts
Analyze diagnostic logs in Log Analytics

Module 15: Manage Azure Cosmos DB at scale with fleets

Understand Azure Cosmos DB fleets
Create and organize a fleet
Configure fleet-level throughput with pools
Monitor a fleet with fleet analytics

Module 16: Implement full-text and vector search in Azure Cosmos DB for NoSQL

Implement full-text search
Design vector data and policies
Configure vector indexes and partitioning
Implement vector search
Keep embeddings current with the change feed

Module 17: Implement hybrid search and optimize AI retrieval in Azure Cosmos DB for NoSQL

Choose a retrieval strategy
Implement hybrid search with RRF
Optimize full-text and vector performance
Isolate search workloads with global secondary indexes

Module 18: Build RAG applications with Azure Cosmos DB for NoSQL

Understand the RAG loop on Azure Cosmos DB
Prepare and embed content for retrieval
Retrieve context and ground responses
Orchestrate a RAG pipeline

Module 19: Design and implement agent memory stores in Azure Cosmos DB

Understand agent memory architectures
Implement conversation state
Implement semantic memory with vector storage
Optimize memory retrieval and context injection
Design memory retention, pruning, and consolidation

Module 20: Implement operational analytics with Microsoft Fabric

Choose Azure Cosmos DB or Cosmos DB in Microsoft Fabric
Configure mirroring into Microsoft Fabric
Monitor mirroring
Query mirrored data with Transact-SQL and Spark

Documentation

Access to Microsoft Learn, Microsoft’s online learning platform, offering interactive resources and educational content to deepen your knowledge and develop your technical skills.

Lab / Exercises

This course provides you with exclusive access to the official Microsoft lab, enabling you to practice your skills in a professional environment.

Exam

This course prepares you to the DP-420: Building AI Applications with Azure Cosmos DB exam.

Complementary Courses

Eligible Funding

ITTA is a partner of a continuing education fund dedicated to temporary workers. This fund can subsidize your training, provided that you are subject to the “Service Provision” collective labor agreement (CCT) and meet certain conditions, including having worked at least 88 hours in the past 12 months.

Additional Information

Azure Cosmos DB: building globally distributed NoSQL applications

Azure Cosmos DB is Microsoft’s fully managed NoSQL database service designed for applications requiring single-digit millisecond response times at global scale. It supports multiple data models including document, key-value, graph and column-family. For developers and architects in Switzerland building cloud-native applications, understanding Cosmos DB’s unique capabilities around global distribution, consistency models and partitioning is essential for designing cost-effective and performant solutions.

What you learn during this 4-day training

The program follows the official Microsoft curriculum, fully refreshed in October 2026. The fundamentals are still there: request units, consistency levels, aggregate modeling, partition keys and hierarchical partition keys, then SDK development with point reads, partial document patch, optimistic concurrency, transactional batches, paginated queries and change feed with Azure Functions. The second half is about AI: vector indexes, hybrid search, RAG pipelines and memory stores for AI agents, plus the Azure Cosmos DB MCP server, Microsoft Entra security, continuous backup, multi-region failover, fleets and mirroring into Microsoft Fabric. Each module includes hands-on labs building real applications on Azure.

Frequently Asked Questions

What are the prerequisites for this DP-420 training?

You should have experience with Azure cloud services, knowledge of JSON data structures and proficiency in C# or Python. Familiarity with NoSQL database concepts is helpful but not required.

Which certification does this training prepare for?

This training prepares you for the DP-420: Building AI Applications with Azure Cosmos DB exam. Since October 6, 2026 it leads to the Microsoft Certified: Azure Cosmos DB AI Developer Associate certification, the new name of the former Azure Cosmos DB Developer Specialty.

When should I choose Cosmos DB over Azure SQL?

Cosmos DB excels for applications requiring global distribution, flexible schemas, massive scale and guaranteed low latency. Azure SQL is better suited for structured relational data with complex transactions. This training helps you understand these trade-offs.