DP-800 study profiles: three SQL backgrounds

For DP-800, the right study route depends on whether your daily work is T-SQL, application code or database operations, because the exam blends all three. Two domains are weighted at 35 to 40% (design and develop, and secure, optimize and deploy) and the AI domain at 25 to 30%. The exam runs 120 minutes with a 700 out of 1000 passing score.

Example profiles, not real people. The profiles below are illustrative planning examples built from the official exam outline. They are not reviews or reports of anyone’s result.

The site’s four-week study plan assumes 8 to 10 hours a week, about 32 to 40 hours, and existing T-SQL. A reasonable planning assumption is to divide those hours near 35% each for the two large domains and 30% for AI, then shift toward your weak area. It is not a guarantee of a pass.

Profile one: the SQL Server or Azure SQL developer

Starting point. This candidate writes CTEs, window functions, procedures and triggers every week and reads execution plans. Schema design and most of the first domain are routine.

Where the time goes. The new items in the AI domain (25 to 30%): external models, embeddings, the vector type and indexes, VECTOR_DISTANCE against VECTOR_SEARCH, hybrid search, and RAG through sp_invoke_external_rest_endpoint. In the deploy half of the second domain, SDK-style SQL Database Projects and Data API builder are rarely part of a day job.

The trap. Studying AI last and cramming it. The AI domain is the smallest in weight, but its vocabulary (ANN against KNN, reciprocal rank fusion, chunking) is new to this profile and the questions are scenario based. The AI capabilities guide covers it.

Suggested order. A quick pass over week one, then weeks three and four of the plan, then security and performance for gaps.

Time budget. A reasonable planning assumption is 14 hours on AI capabilities, 10 on deploy, CI/CD and Data API builder, 6 on a first-domain skim, and the rest on practice.

Profile two: the application developer who uses SQL through an ORM

Starting point. This candidate knows Git, pull requests and pipelines well, and understands REST and JSON. Hand-written T-SQL, locking and security features are thin.

Where the time goes. Design and develop (35 to 40%): the programmability objects, regular expression and JSON functions, graph MATCH queries and TRY...CATCH. Then security and performance: row-level security, Dynamic Data Masking, isolation levels, deadlocks and Query Store.

The trap. Assuming CI/CD knowledge covers the deployment half. The outline asks for database-specific items such as schema drift, data-tier packages and reference data, not generic pipeline skill.

Suggested order. Week one of the plan (T-SQL objects), then week two (security and performance), then the AI topics, with CI/CD as a light review.

Time budget. A reasonable planning assumption is 14 hours on design and develop, 10 on security and performance, 6 on AI topics, and the rest on practice.

Profile three: the DBA who tunes and secures but rarely develops

Starting point. This candidate reads plans, forces them in Query Store, manages permissions and audits, and understands blocking. Writing application-facing T-SQL, JSON and source-controlled schema is less familiar.

Where the time goes. Advanced T-SQL and database objects in the first domain, AI-assisted development with GitHub Copilot and MCP servers, and the whole AI domain. The design and develop guide and secure, optimize and deploy guide separate the areas.

The trap. Over-trusting familiar ground. The exam does not test server patching, backups or high availability, so operational depth earns little beyond tuning and security.

Suggested order. Week one, then week four (AI), then CI/CD, with security and performance as a quick check.

Time budget. A reasonable planning assumption is 12 hours on design and develop, 10 on AI, 6 on deployment and APIs, and the rest on practice.

DP-800 profiles side by side

How to check which DP-800 profile fits before booking

Run the self-checks on the prerequisites page: a window function from memory, OPENJSON, a pull request and a REST call. The exam format page shows the three weights and the October 19, 2026 outline update. Then use the free sample questions and the 20-question practice test to sort errors by domain, and compare with DP-800 versus DP-700 if you are unsure between the two data exams.