
DP-800: Developing AI-Enabled Database Solutions is the exam for Microsoft's SQL AI Developer Associate certification. It is for database developers who build on SQL Server, Azure SQL and SQL databases in Microsoft Fabric. It covers advanced T-SQL (including JSON, regular expression and fuzzy matching functions), security, performance tuning, CI/CD with SQL Database Projects, Data API builder, and AI features inside the database: external models, embeddings, vector and hybrid search, and retrieval-augmented generation (RAG). SQL developers in Hyderabad's banks, product companies and GCCs are now being asked to add AI search and chat to existing databases, and this is the certification that covers those skills.
This guide follows Microsoft's official DP-800 study guide (skills measured as of 12 March 2026) and the certification page. It covers the exam facts, all three skill areas, T-SQL examples for vector search and RAG, a 6-week plan and 15 original practice questions.
Quick facts (Microsoft Learn, November 2026)
| Item | Detail |
|---|---|
| Certification | Microsoft Certified: SQL AI Developer Associate (intermediate level) |
| Exam | DP-800: Developing AI-Enabled Database Solutions |
| Passing score | 700 or greater |
| Price | Based on the country or region where you take the exam |
| Renewal | Every 12 months, through a free online assessment on Microsoft Learn |
| Expected experience | Writing T-SQL and developing databases on Microsoft SQL platforms; CI/CD in GitHub; AI-assisted development tools; embeddings, vectors and models |
| Training | Instructor-led course DP-800T00 (3 days) or free self-paced modules on Microsoft Learn |
Read the official DP-800 study guide and certification page before you book.
Skills measured
| Skill area | Weight | What it covers |
|---|---|---|
| 1. Design and develop database solutions | 35–40% | Tables (columnstore, in-memory, temporal, external, ledger, graph), JSON columns and indexes, constraints, sequences, partitioning; views, functions, stored procedures, triggers; CTEs, window functions, JSON functions, regular expressions (REGEXP_LIKE and others), fuzzy matching (EDIT_DISTANCE, JARO_WINKLER_DISTANCE), graph MATCH queries, error handling; GitHub Copilot and Copilot in Fabric, instruction files and MCP server connections |
| 2. Secure, optimize and deploy database solutions | 35–40% | Always Encrypted, Dynamic Data Masking, Row-Level Security, permissions, passwordless access, auditing, securing model, GraphQL, REST and MCP endpoints; isolation levels, execution plans, DMVs, Query Store, blocking and deadlocks; SQL Database Projects for CI/CD, testing, schema drift and pipeline controls; Data API builder (REST and GraphQL); Azure Monitor; change event streaming, CDC, Change Tracking, Azure Functions SQL trigger and Logic Apps |
| 3. Implement AI capabilities in database solutions | 25–30% | External models; embeddings (which columns, chunking, how to keep them current); full-text, vector and hybrid search; vector data type and indexes; VECTOR_DISTANCE, VECTOR_NORMALIZE, VECTORPROPERTY and VECTOR_SEARCH; ANN vs ENN; reciprocal rank fusion (RRF); RAG with sp_invoke_external_rest_endpoint |
The first two areas are classic SQL development and make up most of the exam. Don't spend all your time on the AI section.
Vector search and RAG in T-SQL
These examples use the vector features in Azure SQL Database and SQL Server 2025. Replace the angle-bracket placeholders with your own Azure OpenAI or Microsoft Foundry resource and deployment names. Some vector features are still in preview, so check the current documentation for exact syntax.
-- 1. A table with a vector column (1536 dimensions for text-embedding-3-small)
CREATE TABLE dbo.Courses (
CourseId int IDENTITY PRIMARY KEY,
Title nvarchar(200) NOT NULL,
Description nvarchar(max) NOT NULL,
Embedding vector(1536) NULL
);
-- 2. Register an external embedding model (uses a database scoped credential)
CREATE EXTERNAL MODEL CourseEmbeddings
WITH (
LOCATION = 'https://<resource>.openai.azure.com/openai/deployments/<embedding-deployment>/embeddings?api-version=<version>',
API_FORMAT = 'Azure OpenAI',
MODEL_TYPE = EMBEDDINGS,
MODEL = 'text-embedding-3-small',
CREDENTIAL = [https://<resource>.openai.azure.com]
);
-- 3. Generate embeddings for existing rows
UPDATE dbo.Courses
SET Embedding = AI_GENERATE_EMBEDDINGS(Description USE MODEL CourseEmbeddings)
WHERE Embedding IS NULL;
-- 4. Exact (ENN) similarity search
DECLARE @q vector(1536) = AI_GENERATE_EMBEDDINGS(N'Kubernetes for beginners' USE MODEL CourseEmbeddings);
SELECT TOP (5) CourseId, Title,
VECTOR_DISTANCE('cosine', Embedding, @q) AS Distance
FROM dbo.Courses
ORDER BY Distance;
ENN vs ANN: VECTOR_DISTANCE with ORDER BY compares every row, which is exact and fine for small tables. For millions of rows you create a vector index and use VECTOR_SEARCH for approximate nearest neighbour (ANN) search, trading a little accuracy for much more speed.
Hybrid search runs a full-text query and a vector query, then merges the two ranked lists with reciprocal rank fusion. Each row scores the sum of 1 / (k + rank) across both lists, where k is a constant, often 60.
-- 5. RAG: send retrieved rows to a chat model
DECLARE @context nvarchar(max) = (
SELECT TOP (3) Title, Description
FROM dbo.Courses
ORDER BY VECTOR_DISTANCE('cosine', Embedding, @q)
FOR JSON PATH);
DECLARE @payload nvarchar(max) = JSON_OBJECT(
'messages': JSON_ARRAY(
JSON_OBJECT('role':'system', 'content':'Answer only from the courses provided.'),
JSON_OBJECT('role':'user', 'content':CONCAT(N'Courses: ', @context, N' Question: Which course suits a fresher?'))));
DECLARE @response nvarchar(max);
EXEC sp_invoke_external_rest_endpoint
@url = N'https://<resource>.openai.azure.com/openai/deployments/<chat-deployment>/chat/completions?api-version=<version>',
@method = 'POST',
@credential = [https://<resource>.openai.azure.com],
@payload = @payload,
@response = @response OUTPUT;
SELECT JSON_VALUE(@response, '$.result.choices[0].message.content') AS Answer;
Security features to practise
-- Dynamic Data Masking: show only the last 4 digits of a phone number
ALTER TABLE dbo.Students
ALTER COLUMN Phone ADD MASKED WITH (FUNCTION = 'partial(0,"XXXXXX",4)');
-- Row-Level Security: each branch sees only its own rows
CREATE FUNCTION Security.fn_Branch(@BranchId int)
RETURNS TABLE WITH SCHEMABINDING AS
RETURN SELECT 1 AS Allowed
WHERE @BranchId = CAST(SESSION_CONTEXT(N'BranchId') AS int);
CREATE SECURITY POLICY Security.BranchFilter
ADD FILTER PREDICATE Security.fn_Branch(BranchId) ON dbo.Enrolments
WITH (STATE = ON);
Know when to use each: Always Encrypted keeps data encrypted even from DBAs; masking hides values from users who lack UNMASK permission but doesn't encrypt; RLS filters rows; object permissions control tables and procedures.
6-week study plan
| Week | Focus | Lab (free Azure SQL Database offer or SQL Server 2025 Developer edition) |
|---|---|---|
| 1 | Tables, constraints, JSON columns, temporal and ledger tables, partitioning | Build a course-enrolment schema with a temporal table |
| 2 | Views, functions, procedures, triggers; CTEs, window functions, JSON, regex and fuzzy matching | Clean messy student names with REGEXP_REPLACE and EDIT_DISTANCE |
| 3 | Security: Always Encrypted, masking, RLS, auditing, passwordless access; Copilot and MCP tools | Apply masking and RLS, then test as different users |
| 4 | Performance: isolation levels, plans, DMVs, Query Store, deadlocks; SQL Database Projects and CI/CD in GitHub | Deploy schema changes with a GitHub Actions pipeline |
| 5 | Data API builder (REST and GraphQL); Azure Monitor; CDC, Change Tracking and change event streaming | Expose a table and a stored procedure through Data API builder |
| 6 | External models, embeddings, chunking, vector and hybrid search, RRF and RAG; practice assessment | Build the course search and RAG example above; book the exam |
15 practice questions with answers
These are original questions written for this guide, not exam questions.
1. Which SQL platforms does DP-800 cover?
SQL Server, Azure SQL, and SQL databases in Microsoft Fabric.
2. Which table type keeps a full history of row changes with period columns?
A system-versioned temporal table.
3. Which table type gives tamper-evidence for data using cryptographic hashes?
A ledger table.
4. Which function returns a scalar value from a JSON string?
JSON_VALUE (OPENJSON returns rows; JSON_QUERY returns an object or array).
5. You need to find customer names that are spelled slightly differently. Which functions help?
Fuzzy string matching functions such as EDIT_DISTANCE, EDIT_DISTANCE_SIMILARITY or JARO_WINKLER_DISTANCE.
6. Masking vs Always Encrypted?
Masking hides values in query results for users without UNMASK permission; Always Encrypted encrypts data so even the database engine and DBAs can't read it in plain text.
7. Two sessions block each other and SQL Server kills one. What happened?
A deadlock; the killed session was chosen as the deadlock victim.
8. Which feature keeps a history of query plans and runtime statistics?
Query Store.
9. How do you detect schema drift between source control and a live database?
Compare the database against the SQL Database Project model, for example with a schema compare in the build or deployment pipeline.
10. What does Data API builder do?
It exposes database tables, views and stored procedures as REST and GraphQL endpoints from a configuration file, without writing API code.
11. Embeddings must update when a product description changes. Name two ways.
For example a table trigger, Change Tracking, CDC, change event streaming, or an Azure Function with a SQL trigger binding.
12. Why split long documents into chunks before embedding?
Embedding models have input limits, and smaller chunks give more precise matches to retrieve.
13. ENN or ANN for a table with 50 million vectors and a strict latency target?
ANN with a vector index, because exact search compares every row.
14. What does reciprocal rank fusion do in hybrid search?
It merges ranked results from full-text and vector search by rank position, so neither score scale dominates.
15. Which system stored procedure calls an external REST endpoint such as a chat model?
sp_invoke_external_rest_endpoint.
Before and after DP-800
- Start with DP-900 and AI-901; brush up with our SQL interview questions.
- Pair it with DP-300 for administration, DP-700 for Fabric data engineering, or AI-103 for AI apps and agents.
- Practise vectors with our vector databases explainer, pgvector RAG tutorial and text-to-SQL agent project.
- Learn it hands-on in our AI, GenAI and Agentic AI course and multi-cloud data and AI engineering course; see live SQL, data and AI developer jobs.
Cloud Soft Solutions in Ameerpet, Hyderabad teaches SQL, data engineering, Azure and GenAI with real-time projects. See the APEX program and all courses.
📞 Call +91 96660 19191 · WhatsApp · Flat No. 513, 5th Floor, Nilagiri Block, Aditya Enclave, Ameerpet, Hyderabad
Frequently Asked Questions
What is the DP-800 exam?
DP-800: Developing AI-Enabled Database Solutions, the exam for Microsoft Certified: SQL AI Developer Associate.
Which platforms does DP-800 cover?
SQL Server, Azure SQL and SQL databases in Microsoft Fabric.
What are the DP-800 skill areas?
Design and develop database solutions (35–40%), secure, optimize and deploy database solutions (35–40%), and implement AI capabilities in database solutions (25–30%).
What is the DP-800 passing score?
700 or greater.
How much does DP-800 cost in India?
Microsoft prices the exam by country or region; check the price for India when you schedule on Microsoft Learn.
Does DP-800 include vector search?
Yes. It covers the vector data type, vector indexes, VECTOR_DISTANCE and VECTOR_SEARCH, ANN vs ENN, hybrid search with reciprocal rank fusion and RAG.



