SQL vector embeddings

Use SQL Server 2025 and Azure SQL vector capabilities to ingest and embed documents.

Project structure

sql-vector-embeddings/ ├── Sql.Vector.Embeddings.AppHost ├── Sql.Vector.Embeddings.BlobUploadConsole ├── Sql.Vector.Embeddings.DatabaseDeploymentService ├── Sql.Vector.Embeddings.Ingestion.Functions ├── Sql.Vector.Embeddings.Ingestion.Core ├── Sql.Vector.Embeddings.QueryConsole ├── Sql.Vector.Embeddings.Data ├── Sql.Vector.Embeddings.Ingestion.Core │ Tests │ ├── Sql.Vector.Embeddings.Data.UnitTests │ ├── Sql.Vector.Embeddings.Ingestion.Core.UnitTests ├── .editorconfig ├── .gitignore ├── README.md └── (other files and directories, e.g., user secrets, settings, and external SQL scripts)

• AppHost - Console app to host the function app locally for testing
• FunctionApp - Azure function app to ingest documents
• DatabaseDeploymentService - DbUp deployment of the SQL database schema and external models
• Data - Data classes
• Ingestion.Core - ingestion application classes

SQL database

If you want to connect to the database with SSMS you can get the connection string in the Aspire portal and paste it into the connection dialog.

Optionally you can set the SQL Server port and sa password by adding these parameters to your secrets. If the parameters aren't set then default values will be generated.

  "Parameters": {
   "SqlServerPassword": "<password>",
   "SqlServerPort": 14331
   }

Call function app

The function app expects an array called uris with an array of uris pointing to pdf files.

There is a test http script in the functions folder that can be used to call it. Alternatively call from a command line using curl:

curl -X POST http://localhost:7034/api/ingest-uris/ -H "Content-Type: application/json" -d '{"uris": ["https://arxiv.org/pdf/1409.0473" ] }'

Aspire hosting and deployment

• See Function integration.

AI model clients

If you are using Ollama and have a GPU, include a parameter OllamaGpuVendor in the AppHost appsettings or secrets. The value can be Nvidia or AMD (or any future values from Aspire.Hosting.OllamaGpuVendor). This is added via an extension WithGPUSupportIfVendorParameterProvided and should match your system.

  "Parameters": {
   "OllamaGpuVendor": "Nvidia"
   }

Arxiv data

The ingestion function uses the arXiv API to load document metadata and then to download documents.